The Role of Preoperative Laboratory Test Indicators in Predicting Thrombosis Risk in Elderly Hip Fracture Patients: A Random Forest Approach

The Role of Preoperative Laboratory Test Indicators in Predicting Thrombosis Risk in Elderly Hip Fracture Patients: A Random Forest Approach | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article The Role of Preoperative Laboratory Test Indicators in Predicting Thrombosis Risk in Elderly Hip Fracture Patients: A Random Forest Approach Xuehai Jia, Hao Liu, Changyong Shen, Yi Deng, Kerui Zhang, Litai Ma, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6339181/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background: Deep vein thrombosis (DVT) poses a common and critical risk for mortality in elderly hip fracture (HF) patients. Venous angiography and ultrasound examinations serve as crucial diagnostic tools but pose challenges in cases with prevalent complications. The extensive training period for technical personnel, coupled with the rapid advancements in machine learning, prompts our research to harness the potential of the random forest algorithm. Our aim is to construct a predictive model that evaluates the risk of thrombosis formation in elderly hip fracture patients upon admission. Methods: We conducted a retrospective evaluation of 448 elderly HF patients who received surgical treatment between May 2021 and November 2023. The study cohort was partitioned into training and test datasets, maintaining a 70:30 ratio. Leveraging the Random Forest algorithm, we developed a streamlined predictive model. Results: Eleven important variables, namely ALB, A/G, PLT, Fib, D-dimer, CREA, Pi, GFR, FDP, fracture onset to surgery time, and cardiopulmonary diseases, were screened based on Random Forest features. In the training set, AUC, the 95% Confidence Interval (CI), Sensitivity, Specificity, Precision, Accuracy, and Balanced Accuracy stand at 1.000, (0.9883, 1), 1.000, 1.000, 1.000, 1.000, and 1.000, respectively. For the test set, AUC, the 95% CI, Sensitivity, Specificity, Precision, Accuracy, and Balanced Accuracy are 0.899, (0.7875, 0.9124), 0.6061, 0.9406, 0.7692, 0.8582, and 0.7733, respectively. Conclusions: A random forest prediction model was developed to anticipate the occurrence of preoperative lower extremity DVT in elderly HF patients. This model demonstrated superior accuracy compared to the logistic regression model. Key preoperative laboratory test indicators proved valuable as variables in the prediction process. Biological sciences/Physiology/Bone Health sciences/Medical research/Outcomes research machine learning random forest HF deep vein thrombosis prediction model Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Introduction With the global aging population, osteoporosis and related conditions (such as increased fall risk) have led to a rise in the incidence of hip fractures (HF) in elderly individuals[ 1 – 4 ]. Hip fractures are not only a common and severe health issue among the elderly, but they also significantly increase the burden on healthcare systems[ 5 ]. In addition to the direct physical consequences, elderly HF patients face a higher risk of complications, especially deep vein thrombosis (DVT) [ 6 ]. After hip fractures, prolonged immobility, surgical intervention, and other factors increase the likelihood of DVT formation, which can escalate into life-threatening conditions such as pulmonary embolism (PE) [ 7 , 8 ]. DVT is a common complication in HF patients, and the postoperative mortality rate in these patients is often exacerbated by DVT, making it a key issue in managing elderly patients with hip fractures[ 9 ]. Clinicians are increasingly concerned about preoperative DVT diagnosis and prevention, necessitating effective strategies for preoperative DVT risk assessment. Early identification of elderly hip fracture patients at high risk for preoperative or admission DVT formation is crucial. Timely and accurate diagnosis and treatment are essential to mitigate the risks of massive intraoperative pulmonary embolism, postoperative mortality, and morbidity in hip fracture patients[ 10 , 11 ]. However, blood tests for known proinflammatory and prothrombotic states have limitations, and relying on a single indicator to accurately predict DVT formation upon admission is challenging[ 12 ]. Limited studies have shown that risk factors associated with preoperative DVT include postmenopausal women, previous venous thromboembolism, chronic kidney disease, delayed surgery, chronic obstructive pulmonary disease (COPD), prolonged immobilization, low mean platelet volume, abnormal serum pH, and elevated D-dimer levels[ 6 , 13 – 15 ]. These risk factors are identified through single-factor analyses, highlighting the need to construct predictive models for DVT by identifying additional risk factors. As DVT diagnosis requires skilled ultrasound technicians, the training period for these specialized examiners is extended. We attempt to construct a predictive model for forecasting preoperative DVT formation by utilizing patient medical histories and characteristics available upon admission, along with necessary preoperative test indicators. The random forest predictive model can iteratively learn from new patient data, enhancing its adaptability to evolving medical information. Nowadays, machine learning, particularly the random forest algorithm—a subset of machine learning algorithms—has emerged as a promising tool for predicting preoperative DVT in elderly hip fracture patients[ 16 ]. The random forest algorithm, known for constructing mathematical models based on sample data, has demonstrated efficacy in predicting outcomes in end-stage renal disease and type A aortic dissection (TAAD) [ 17 , 18 ]. In this study, the predictive model using the random forest algorithm aims to uncover key risk factors and enhance the accuracy of preoperative DVT diagnosis in elderly hip fracture patients. By utilizing diverse data points, including the patient's physical condition, comorbidities, and clinical parameters, this research strives to construct a machine learning model tailored to the specific dynamics of elderly hip fracture cases. Figure 1 illustrates the step-by-step process of building the preoperative DVT prediction model in this study. Ultimately, the predictive modeling facilitated by the random forest algorithm lays the groundwork for improved preoperative care and a better prognosis for this aging hip fracture patient population. Methods General Information and Data Collection Data for the two groups in this study (DVT group and non-thrombosis group) were collected from patients undergoing surgical intervention for hip fractures (HF) at West China Hospital of Sichuan University between May 2021 and November 2023. The study received approval from the Ethics Committee of West China Hospital of Sichuan University. The inclusion criteria for this study were as follows: (1) Age ≥ 60 years. (2) Patients who underwent surgical intervention during hospitalization. (3) Patients diagnosed with HF through physical examination combined with medical imaging (x-ray or computed tomography). (4) Complete clinical data. The study excluded patients with the following characteristics: (1) Secondary fractures. (2) Pre-existing hip fractures. (3) Pathologic fractures. (4) Multiple fractures or polytrauma. (5) Open fractures. (6) Old fractures (> 7 days). (7) Patients with poor compliance, such as those with mental illness. (8) Those undergoing conservative treatment. (9) Patients in the acute stage of cardiovascular disease. (10) Individuals without preoperative lower extremity vascular ultrasonography. (11) Patients with a history of DVT before the fracture. (12) Those involved in high-energy trauma mechanisms. (13) Cases with incomplete clinical information. In total, case data were collected from 448 elderly hip fracture patients. All collected data were analyzed anonymously, with no personal identifiers or personal privacy involved, and stored in an electronic database in Excel format. Subsequently, all cases were randomly divided into training and test datasets in a 70:30 ratio, assigning 314 patients to the training group and 134 patients to the test group. Clinical information for all patients was acquired from four distinct sources, encompassing demographics, chronic comorbidities, fracture-related data, and blood laboratory tests. Demographic data included patient-specific details such as gender, age, height, weight, and the calculation of Body Mass Index (BMI). Chronic comorbidities included factors related to long-standing health conditions, including alcohol consumption, smoking habits, hypertension, diabetes, and cardiopulmonary diseases. In this research, variables related to cardiopulmonary diseases primarily included Chronic Obstructive Pulmonary Disease (COPD), pulmonary infections, cardiac arrhythmias, clinically symptomatic heart failure, and coronary heart disease. Fracture-related data involved information regarding the classification of hip fractures and the duration between the occurrence of the fracture and the subsequent surgical intervention. Laboratory blood test indicators comprised the results of necessary preoperative tests, providing insights into the patient's physiological status. In this study, preoperative essential blood laboratory tests included blood cell analysis, a comprehensive coagulation panel, and biochemical blood analysis. We collected data on 63 blood indicators and, through single-factor analysis, random combinations, and correlation analysis among them, ultimately selected 22 blood indicators for inclusion in the predictive model as continuous variables. This study aimed to predict the occurrence of deep vein thrombosis (DVT) in hip fractures by gathering patient history, demographic characteristics, and readily available blood test results after admission. Laboratory biomarkers covered a comprehensive range, such as prothrombin time (PT), International Normalized Ratio (INR), fibrinogen (Fib), fibrin and fibrinogen degradation products (FDP), D-dimer, platelet count (PLT), white blood cell count (WBC), total bilirubin (TB), alanine aminotransferase (ALT), aspartate aminotransferase (AST), albumin (ALB), globulin (Glb), albumin-to-globulin ratio (A/G), creatinine (CREA), estimated glomerular filtration rate (GFR), triglyceride (TG), cholesterol (TC), creatine kinase (CK), lactate dehydrogenase (LDH), anion gap (AG), calcium (Ca), magnesium (Mg), and inorganic phosphorus (Pi). If patients underwent multiple hematological examinations after admission, we selected the first examination for analysis. The variables were also scrutinized for correlation. Diagnosis and management of DVT Deep vein thrombosis (DVT) of the lower extremities is a serious vascular condition characterized by the formation of blood clots in the deep veins of the legs. If left untreated, it can lead to life-threatening complications, such as pulmonary embolism. Early detection and management are essential. The diagnosis of DVT typically involves a comprehensive approach, including a thorough physical examination and a detailed assessment of the patient's symptoms. Ultrasound of the proximal leg veins plays a critical role in confirming the presence of DVT. This imaging technique is widely recognized and endorsed in medical guidelines, such as those provided by the American Society of Hematology[ 19 ]. Treatment for DVT primarily consists of anticoagulant medications, which help prevent clot growth and reduce the risk of complications. Effective management underscores the importance of accurate diagnosis and early intervention, both of which significantly improve patient outcomes by reducing the likelihood of severe complications[ 20 ]. Statistical analysis For continuous variables not conforming to a normal distribution, statistical analysis utilized nonparametric tests, and descriptive statistics employed medians (quartiles). Normal distribution conformity for continuous variables was assessed using the Kolmogorov-Smirnov method. For instance, non-paired t-tests or Wilcoxon rank-sum tests were used to compare differences between the DVT and non-DVT groups for variables like BMI and ALB in Table 2 , as well as A/G and Ca variables in Table 3 . Furthermore, based on the specific circumstances, the application of Pearson's chi-square test or Fisher's exact test was considered. Descriptive statistics, using means with positive and negative standard deviations, facilitated comparisons between groups using the independent samples t-test. Categorical variables were presented as numbers and percentages (%), and group comparisons employed the chi-square test or Fisher's exact test, as applicable. The Random Forest algorithm is a powerful machine learning method known for its accuracy and robustness in handling classification and regression tasks. Proposed by Leo Breiman in 2001, this algorithm operates by constructing a large number of decision trees. It combines the votes from each decision tree to perform effective classification or regression analyses. Each decision tree is trained on a different subset of data, and their outputs are aggregated to enhance predictive performance. The algorithm effectively captures complex relationships within the data, reducing overfitting concerns and improving generalization capabilities. Notably, it excels at reducing the risk of overfitting, effectively managing large datasets with numerous features, and highlighting the importance of individual features. The specified configuration involved the use of 500 decision trees. Data Preparation Gender, alcohol consumption, smoking, hypertension, diabetes, classification of hip fractures, and cardiopulmonary disease were entered as dichotomous variables. Age, BMI, fracture onset to surgery time, and laboratory tests were included as continuous variables in the Random Forest program. Bootstrap Sampling Bootstrapping and resampling/Bagging methods were used to randomly select a subset of samples for training multiple decision trees. Feature Randomization A subset of m features was randomly chosen for each tree split (the Gini coefficient calculates the optimal segmentation method for m features, where m is less than the total number of features). Input variables in the risk prediction model were ranked based on the average rate of decrease in accuracy and the average rate of decrease in the Gini coefficient. Decision Tree Construction Individual decision trees were constructed using sampled data and features. Voting Mechanism Predictions from all trees were aggregated to generate the final output for classification or regression. Evaluating the Model The model's performance was assessed using metrics such as precision or mean square error. In the Bagging approach, each model's training set omitted (excluded) a number of data points, constituting the Out-of-Bag (OOB) samples for each model. The OOB data was not used for training models but served as a test set. The overall estimation and generalization ability of the model's predictive performance were robustly assessed by averaging the OOB scores of all models and the OOB error. Additionally, Lasso regression was employed, considering the prominent feature selection ability of the Least Absolute Shrinkage and Selection Operator (Lasso) and the regression analysis ability of regularization. The prediction model, constructed through logistic regression, was validated on the training and test datasets by comparing the results to those of the Random Forest algorithm. Logistic regression, a linear regression technique based on the log probability ratio of the response variable, was used. Finally, the Area Under the Curve (AUC) for continuous variables and dichotomous variables was calculated. The AUC values from the random forest algorithm and traditional logistic regression were compared. The Kappa statistic and F-measure were used to test the reliability of the model. Other assessment tools included 95% confidence interval (CI), sensitivity, specificity, precision, accuracy, and balanced accuracy. The flow of the prediction model for lower extremity deep vein thrombosis in this study is depicted in Fig. 1 . All statistical analyses in this study were conducted using RStudio (Version 2023.09.1 Build 494© 2009–2023 Posit Software, PBC, "Desert Sunflower" Release [cd7011dc, 2023-10-16]) for Windows. The R software version used was 4.3.2. RStudio, along with its library packages Boruta (CRAN.R-project.org/package = Boruta), randomForest (CRAN.R-project.org/package = randomForest), and caret (CRAN.R-project.org/package = caret), were used for constructing random forest models. Additionally, Lasso regression was performed using RStudio and the glmnet library package. Statistical significance was determined by a p-value of < 0.05. Results Patient Characteristics In this study, 24.55% (110/448) of the enrolled patients were elderly individuals with hip fractures (HF) and preoperative lower extremity deep vein thrombosis (DVT). Figure 1 illustrates the step-by-step process of building the preoperative DVT prediction model in this study. The median age of the participants was 81 (71, 87), and there was a female predominance (63%). Table 1 displays the demographics, medical history, and laboratory test results of all enrolled patients. In the baseline comparison, statistically significant differences (P < 0.05) in clinical variables were evident between the DVT and non-DVT groups. These differences included age (P = 0.047), Fib (P < 0.001), FDP (P < 0.001), D-dimer (P < 0.001), PLT (P < 0.001), cardiopulmonary disease (P = 0.009), and fracture onset to surgery time (P < 0.001). Moreover, statistically significant variations were observed in WBC (P = 0.043), A/G (P < 0.001), ALB (P < 0.001), GFR (P < 0.001), LDH (P < 0.001), and Pi (P < 0.001). The AUC for each variable in the univariate analysis is depicted in Fig. 2 . Further, D-dimer levels were 6.24 (3.91, 11.35) µg/mL in the non-DVT group and 9.38 (6.85, 15.01) µg/mL in the DVT group. PLT was 161 (123, 201) × 10^3/µL in the non-DVT group and 213 (165, 266) × 10^3/µL in the DVT group. ALB levels were 37.2 (31.9, 42.4) g/dL in patients without DVT and 31.75 (27.45, 35.98) g/dL in those with DVT. Fib levels were 3.33 (2.22, 4.26) g/L in the non-DVT group and 3.76 (3.16, 4.97) g/L in the DVT group. Additionally, fracture onset to surgery time was 4.2 (3.3, 5.1) days in the non-DVT group and 4.9 (4.2, 5.8) days in the DVT group. FDP was 10.1 (4.9, 22.48) µg/mL in the non-DVT group and 13.75 (8.45, 25.65) µg/mL in the DVT group. Pi was 1.04 (0.89, 1.26) mg/dL in the non-DVT group and 1.25 (1.03, 1.53) mg/dL in the DVT group. Significant variations between the non-DVT and DVT groups also included A/G, with respective values of 1.38 (1.21, 1.6) and 1.2 (1.06, 1.4); GFR, with values of 83.69 (64.97, 96.11) in the non-DVT group and 70.38 (51.98, 82.62) in the DVT group; and LDH, with values of 208 (181.25, 240) in the non-DVT group and 225 (192.25, 275.5) in the DVT group. The differences in the distribution of these laboratory indices between the DVT and non-DVT groups were statistically significant (P < 0.001). Figure 3 displays the data visualization of the distribution of continuous variables in the DVT and non-DVT groups. Data visualization of dichotomous variables in the non-DVT and DVT groups is presented in Fig. 4 . Comorbidities among all enrolled patients included hypertension in 203 individuals, diabetes in 123 individuals, a history of cardiopulmonary disease in 144 individuals, a history of smoking in 173 individuals, and a history of drinking in 120 individuals. The results of the correlation analysis and significance tests for each variable are illustrated in Fig. 5 . Subsequently, all cases were randomly allocated to a training dataset (Table 2 ) and a test dataset (Table 3 ) in a 70:30 ratio for further analysis of their baseline characteristics. Table 1 Demographic and clinical characteristics of study participants (Total). Variables Total (n = 448) No-DVT group DVT group p (n = 338,75%) (n = 110,25%) Demographics Sex, n (%) 0.998 Female 283 (63) 213 (63) 70 (64) Male 165 (37) 125 (37) 40 (36) Age, Median (Q1, Q3) 81 (71, 87) 81 (71, 86) 83 (74, 88) 0.047 BMI, Median (Q1, Q3) 21.92 (19.78, 24.29) 22 (19.98, 24.34) 21.42 (18.86, 24.18) 0.161 Medical history Hypertension, n (%) 0.715 No 245 (55) 187 (55) 58 (53) Yse 203 (45) 151 (45) 52 (47) Diabetes, n (%) 0.248 No 325 (73) 240 (71) 85 (77) Yse 123 (27) 98 (29) 25 (23) Smoking, n (%) 0.826 No 275 (61) 206 (61) 69 (63) Yse 173 (39) 132 (39) 41 (37) Drinking, n (%) 0.811 No 328 (73) 246 (73) 82 (75) Yse 120 (27) 92 (27) 28 (25) Cardiopulmonary diseases, n (%) 0.009 No 304 (68) 241 (71) 63 (57) Yse 144 (32) 97 (29) 47 (43) Hip fractures Classification, n (%) 0.196 Extra-articular fracture 247 (55) 180 (53) 67 (61) Intra-articular fracture 201 (45) 158 (47) 43 (39) Fracture onset to surgery time (days) 4.4 (3.4, 5.2) 4.2 (3.3, 5.1) 4.9 (4.2, 5.8) < 0.001 Median (Q1,Q3) Laboratory tests PT (s), Median (Q1, Q3) 11.7 (11.2, 12.4) 11.7 (11.2, 12.4) 11.8 (11.3, 12.4) 0.751 APTT (s), Median (Q1, Q3) 28.3 (26.17, 30.5) 28.3 (26.13, 30.58) 28.1 (26.3, 30.23) 0.42 INR, Median (Q1, Q3) 1.07 (1.01, 1.12) 1.07 (1.01, 1.12) 1.07 (1.01, 1.12) 0.993 Fib (g/L), Median (Q1, Q3) 3.47 (2.5, 4.47) 3.33 (2.22, 4.26) 3.76 (3.16, 4.97) < 0.001 FDP (mg/mL), Median (Q1, Q3) 11.3 (5.7, 22.95) 10.1 (4.9, 22.48) 13.75 (8.45, 25.65) < 0.001 D-dimer (mg/l FEU), Median (Q1, Q3) 7.12 (4.23, 12.2) 6.24 (3.91, 11.35) 9.38 (6.85, 15.01) < 0.001 RBC (×10^12/L), Median (Q1, Q3) 3.59 (3.01, 4.16) 3.64 (3.08, 4.17) 3.42 (2.94, 4.04) 0.12 PLT (×10^9/L), Median (Q1, Q3) 171 (133.75, 223.75) 161 (123, 201) 213 (165, 266) < 0.001 WBC (×10^9/L), Median (Q1, Q3) 8.54 (6.69, 11.14) 8.46 (6.58, 10.56) 8.82 (7, 12.46) 0.043 ALT (U/L), Median (Q1, Q3) 14 (10, 21) 14 (11, 21) 14 (10, 22) 0.954 AST (U/L), Median (Q1, Q3) 20.5 (17, 26) 20 (17, 25) 21 (17, 31) 0.309 ALB (g/L), Median (Q1, Q3) 35.9 (30.48, 40.8) 37.2 (31.9, 42.4) 31.75 (27.45, 35.98) < 0.001 Glb (g/L), Median (Q1, Q3) 24.7 (21.8, 27.7) 24.7 (21.92, 27.67) 24.1 (21.8, 28.12) 0.544 A/G, Median (Q1, Q3) 1.35 (1.18, 1.56) 1.38 (1.21, 1.6) 1.2 (1.06, 1.4) < 0.001 CREA (umol/L), Median (Q1, Q3) 92 (76, 123) 90.5 (76, 117.5) 93 (73, 165.5) 0.409 GFR (mL/min/1.73m²), Median (Q1, Q3) 79.47 (61.44, 93.64) 83.69 (64.97, 96.11) 70.38 (51.98, 82.62) < 0.001 TG (mmol/L), Median (Q1, Q3) 1.05 (0.82, 1.38) 1.06 (0.84, 1.37) 1.01 (0.71, 1.43) 0.258 TC (mmol/L), Median (Q1, Q3) 3.46 (2.88, 3.92) 3.46 (2.87, 3.95) 3.47 (2.88, 3.92) 0.852 CK (U/L), Median (Q1, Q3) 95 (55.75, 207.5) 91 (54.25, 179.75) 114 (58, 264.5) 0.057 LDH (U/L), Median (Q1, Q3) 212 (185, 249.25) 208 (181.25, 240) 225 (192.25, 275.5) < 0.001 Ca(mmol/L), Median (Q1, Q3) 2.14 (2.06, 2.23) 2.14 (2.06, 2.23) 2.14 (2.06, 2.23) 0.795 Pi(mmol/L), Median (Q1, Q3) 1.08 (0.91, 1.35) 1.04 (0.89, 1.26) 1.25 (1.03, 1.53) < 0.001 Median (Q1, Q3): Continuous variables that did not fit the normal distribution used the median (quartiles). Mean ± SD: Continuous variables that fit the normal distribution were tested to obtain the mean (plus or minus standard deviation, SD) using the Kolmogorov-Smirnov method. n (%): Categorical variables expressed as numbers and percentages (%). Table 2 Demographic and clinical characteristics of training group participants. Variables Trains (n = 314) No-DVT group DVT group p (n = 237,75%) (n = 77,25%) Demographics Sex, n (%) 0.741 Female 201 (64) 150 (63) 51 (66) Male 113 (36) 87 (37) 26 (34) Age, Median (Q1, Q3) 81 (72, 88) 81 (71, 87) 83 (74, 89) 0.083 BMI, Mean ± SD 22.16 ± 3.5 22.22 ± 3.35 21.98 ± 3.96 0.638 Medical history Hypertension, n (%) 1 No 176 (56) 133 (56) 43 (56) Yse 138 (44) 104 (44) 34 (44) Diabetes, n (%) 0.76 No 222 (71) 166 (70) 56 (73) Yse 92 (29) 71 (30) 21 (27) Smoking, n (%) 0.393 No 193 (61) 142 (60) 51 (66) Yse 121 (39) 95 (40) 26 (34) Drinking, n (%) 0.736 No 234 (75) 175 (74) 59 (77) Yse 80 (25) 62 (26) 18 (23) Cardiopulmonary diseases, n (%) 0.017 No 216 (69) 172 (73) 44 (57) Yse 98 (31) 65 (27) 33 (43) Hip fractures Classification, n (%) 0.316 Extra-articular fracture 170 (54) 124 (52) 46 (60) Intra-articular fracture 144 (46) 113 (48) 31 (40) Fracture onset to surgery time (days) 4.4 (3.4, 5.2) 4.2 (3.2, 5) 5 (4.3, 5.8) < 0.001 Median (Q1,Q3) Laboratory tests PT (s), Median (Q1, Q3) 11.8 (11.3, 12.5) 11.8 (11.3, 12.5) 11.8 (11.2, 12.4) 0.665 APTT (s), Median (Q1, Q3) 28.15 (26, 30.28) 28.2 (26, 30.2) 28.1 (26, 30.6) 0.919 INR, Median (Q1, Q3) 1.07 (1, 1.12) 1.06 (1, 1.12) 1.07 (1, 1.12) 0.733 Fib (g/L), Median (Q1, Q3) 3.43 (2.48, 4.41) 3.32 (2.18, 4.2) 3.75 (3.14, 4.87) < 0.001 FDP (mg/mL), Median (Q1, Q3) 11 (5.8, 22.58) 9 (5, 21.4) 15.3 (8.7, 24.2) 0.001 D-dimer (mg/l FEU), Median (Q1, Q3) 7.32 (4.23, 12.14) 6.26 (3.85, 11.36) 9.77 (7.29, 13.54) < 0.001 RBC (×10^12/L), Median (Q1, Q3) 3.58 (3.03, 4.15) 3.65 (3.13, 4.17) 3.34 (2.89, 3.97) 0.029 PLT (×10^9/L), Median (Q1, Q3) 169 (135, 231) 161 (123, 198) 215 (160, 259) < 0.001 WBC (×10^9/L), Median (Q1, Q3) 8.68 (6.71, 11.33) 8.52 (6.59, 10.99) 8.84 (7.46, 12.41) 0.124 ALT (U/L), Median (Q1, Q3) 14 (11, 21) 15 (11, 21) 14 (10, 24) 0.838 AST (U/L), Median (Q1, Q3) 21 (17, 27) 21 (17, 25) 21 (17, 37) 0.164 ALB (g/L), Mean ± SD 35.47 ± 7.72 36.9 ± 7.69 31.05 ± 6 < 0.001 Glb (g/L), Median (Q1, Q3) 24.7 (21.8, 27.9) 24.7 (21.8, 27.8) 24.1 (21.8, 29.3) 0.684 A/G, Median (Q1, Q3) 1.34 (1.17, 1.56) 1.37 (1.21, 1.6) 1.19 (1.07, 1.4) < 0.001 CREA (umol/L), Median (Q1, Q3) 92.5 (76, 126) 89 (76, 116) 96 (73, 167) 0.292 GFR (mL/min/1.73m²), Median (Q1, Q3) 78.75 (60.43, 93.08) 83.19 (63.02, 95.47) 69.19 (51.89, 80) < 0.001 TG (mmol/L), Median (Q1, Q3) 1.09 (0.8, 1.43) 1.13 (0.84, 1.43) 0.97 (0.69, 1.38) 0.054 TC (mmol/L), Median (Q1, Q3) 3.46 (2.88, 3.88) 3.48 (2.88, 3.89) 3.45 (2.82, 3.78) 0.718 CK (U/L), Median (Q1, Q3) 101 (58, 234.25) 101 (57, 197) 121 (59, 287) 0.109 LDH (U/L), Median (Q1, Q3) 214.5 (186.25, 251) 208 (182, 244) 225 (196, 270) 0.006 Ca(mmol/L), Median (Q1, Q3) 2.14 (2.07, 2.23) 2.14 (2.07, 2.23) 2.14 (2.06, 2.23) 0.772 Pi(mmol/L), Median (Q1, Q3) 1.08 (0.89, 1.34) 1.05 (0.89, 1.27) 1.23 (1.03, 1.46) < 0.001 Median (Q1, Q3): Continuous variables that did not fit the normal distribution used the median (quartiles). Mean ± SD: Continuous variables that fit the normal distribution were tested to obtain the mean (plus or minus standard deviation, SD) using the Kolmogorov-Smirnov method. n (%): Categorical variables expressed as numbers and percentages (%). Table 3 Demographic and clinical characteristics of test group participants. Variables Test (n = 134) No-DVT group DVT group p (n = 101,75%) (n = 33,25%) Demographics Sex, n (%) 0.775 Female 82 (61) 63 (62) 19 (58) Male 52 (39) 38 (38) 14 (42) Age, Median (Q1, Q3) 80 (70.25, 85) 79 (70, 84) 82 (71, 86) 0.28 BMI, Median (Q1, Q3) 21.48 (19.05, 23.78) 21.83 (20.03, 24) 20.54 (18.36, 22.94) 0.175 Medical history Hypertension, n (%) 0.549 No 69 (51) 54 (53) 15 (45) Yse 65 (49) 47 (47) 18 (55) Diabetes, n (%) 0.136 No 103 (77) 74 (73) 29 (88) Yse 31 (23) 27 (27) 4 (12) Smoking, n (%) 0.486 No 82 (61) 64 (63) 18 (55) Yse 52 (39) 37 (37) 15 (45) Drinking, n (%) 1 No 94 (70) 71 (70) 23 (70) Yse 40 (30) 30 (30) 10 (30) Cardiopulmonary diseases, n (%) 0.359 No 88 (66) 69 (68) 19 (58) Yse 46 (34) 32 (32) 14 (42) Hip fractures Classification, n (%) 0.533 Extra-articular fracture 77 (57) 56 (55) 21 (64) Intra-articular fracture 57 (43) 45 (45) 12 (36) Fracture onset to surgery time (days) 4.3 (3.43, 5.27) 4.2 (3.4, 5.3) 4.7 (4.2, 5.2) < 0.001 Median (Q1,Q3) Laboratory tests PT (s), Median (Q1, Q3) 11.5 (11.1, 12.3) 11.4 (11.1, 12.3) 11.6 (11.3, 12.3) 0.153 APTT (s), Median (Q1, Q3) 28.7 (26.72, 31.15) 28.9 (26.9, 31.5) 27.9 (26.7, 29.5) 0.081 INR, Median (Q1, Q3) 1.07 (1.03, 1.13) 1.07 (1.03, 1.13) 1.06 (1.01, 1.13) 0.576 Fib (g/L), Median (Q1, Q3) 3.49 (2.55, 4.61) 3.34 (2.47, 4.35) 3.81 (3.17, 5.17) 0.02 FDP (mg/mL), Median (Q1, Q3) 11.9 (5.62, 24.53) 11.3 (4.7, 23.8) 12.2 (8.2, 25.8) 0.184 D-dimer (mg/l FEU), Median (Q1, Q3) 6.76 (4.31, 12.58) 6.16 (4.03, 10.36) 8.35 (6.69, 19.42) < 0.001 RBC (×10^12/L), Median (Q1, Q3) 3.62 (3, 4.22) 3.61 (3.05, 4.21) 3.64 (3, 4.28) 0.668 PLT (×10^9/L), Median (Q1, Q3) 172 (133.25, 212.5) 160 (122, 201) 200 (172, 278) < 0.001 WBC (×10^9/L), Median (Q1, Q3) 8.42 (6.6, 10.71) 8.32 (6.48, 10.23) 8.73 (6.78, 12.7) 0.184 ALT (U/L), Median (Q1, Q3) 13 (10, 19) 13 (10, 19) 14 (10, 18) 0.8 AST (U/L), Median (Q1, Q3) 20 (16, 25) 20 (16, 25) 20 (17, 23) 0.784 ALB (g/L), Median (Q1, Q3) 37.05 (30.55, 41.38) 38.3 (32.3, 43.5) 32.7 (27.7, 36.2) < 0.001 Glb (g/L), Median (Q1, Q3) 24.65 (22, 27.17) 24.7 (22, 27.2) 24.2 (21.6, 26.9) 0.657 A/G, Mean ± SD 1.37 ± 0.33 1.41 ± 0.32 1.22 ± 0.31 0.004 CREA (umol/L), Median (Q1, Q3) 87 (69, 123) 91 (72, 119) 79 (69, 153) 0.92 GFR (mL/min/1.73m²), Median (Q1, Q3) 82.46 (65.51, 95.64) 87.09 (66.79, 97.86) 74 (52.23, 88.89) 0.022 TG (mmol/L), Median (Q1, Q3) 1.04 (0.85, 1.3) 1.02 (0.85, 1.28) 1.21 (0.85, 1.52) 0.242 TC (mmol/L), Median (Q1, Q3) 3.46 (2.87, 4.28) 3.45 (2.8, 4.18) 3.52 (3.11, 4.31) 0.303 CK (U/L), Median (Q1, Q3) 85.5 (48.25, 143) 83 (48, 135) 101 (51, 206) 0.271 LDH (U/L), Median (Q1, Q3) 209.5 (181.25, 246.75) 209 (181, 236) 217 (185, 289) 0.057 Ca(mmol/L), Median (Q1, Q3) 2.13 ± 0.13 2.14 ± 0.13 2.13 ± 0.15 0.694 Pi(mmol/L), Mean ± SD 1.06 (0.95, 1.4) 1.03 (0.93, 1.25) 1.36 (1.06, 1.56) 0.001 Median (Q1, Q3): Continuous variables that did not fit the normal distribution used the median (quartiles). Mean ± SD: Continuous variables that fit the normal distribution were tested to obtain the mean (plus or minus standard deviation, SD) using the Kolmogorov-Smirnov method. n (%):Categorical variables expressed as numbers and percentages (%). Feature Screening In this study, the random forest prediction model was employed to meticulously examine each variable for its potential to predict the occurrence of preoperative deep vein thrombosis (DVT) in elderly patients with hip fractures (HF). The process began with random forest-based feature screening, as depicted in Figs. 6 A and B, which illustrated both the process and outcomes of feature selection via the random forest algorithm. From these results, 11 key attributes were identified as crucial for predicting the target variable (DVT). These attributes are ALB (albumin), D-dimer, PLT (platelet count), Fib (fibrinogen), A/G (albumin-to-globulin ratio), Pi (inorganic phosphorus), CREA (creatinine), GFR (estimated glomerular filtration rate), FDP (fibrin and fibrinogen degradation products), cardiopulmonary disease, and fracture onset to surgery time. Additionally, three tentative attributes—AST (aspartate aminotransferase), sex, LDH (lactate dehydrogenase), and TG (triglyceride)—were identified and require further investigation due to uncertain importance. By utilizing Boruta analysis for feature selection, the model was streamlined, focusing on the most relevant variables to enhance prediction accuracy and efficiency. We then identified the feature selection results of the Random Forest algorithm through the Lasso regression model. The Lasso coefficient distributions of the features and the optimal penalty coefficients (λ) are shown in Fig. 7A-B. Unsurprisingly, due to the strong contraction and regularized regression analysis ability of Lasso regression, only 9 variables were left in the Lasso regression model associated with the occurrence of preoperative DVT in elderly HF patients. This set of variables completely overlapped with the important attribute variables identified in Random Forest. Subsequently, we leveraged the Lasso regression model to validate the feature selection outcomes derived from the Random Forest algorithm. As depicted in Fig. 7A-B, the distributions of Lasso coefficients for the features and the optimal penalty coefficients (λ) are presented. Given the inherent capacity of Lasso regression to impose strong constraints and conduct regularized regression analysis, it comes as no surprise that only nine variables remained in the Lasso regression model, which are associated with the occurrence of preoperative DVT in elderly patients with hip fractures (HF). Remarkably, this set of variables fully coincided with the important attribute variables identified through Random Forest, thereby reinforcing the robustness and reliability of the selected features.Figure 7. (A). Lasso coefficient profiles created from log (lambda) series. (B). Ten-fold cross-validation of predictor variables with minimum criterion (left) and minimum plus 1 standard error (right) to determine optimal tuning parameters (lambda). Random Forest Prediction Model Construction and Validation Following the identification of the 11 crucial risk factors through the Random Forest algorithm, we proceeded to construct a risk prediction model. The model was built using these key variables, and the results indicated that a random forest classification model with a 500-class decision tree, which attempts four variables at each segmentation, performed effectively. The out-of-bag (OOB) error rate was calculated to be 13.69%, signifying a relatively small generalization error and suggesting that the model has good predictive performance. The Receiver Operating Characteristic (ROC) curves for continuous variables, predictive models constructed by the random forest algorithm, and traditional logistic regression were visualized in both the training dataset (Fig. 8 ) and the test dataset (Fig. 9 ). The area under the curve (AUC) for the prediction model based on the random forest algorithm was 1.000 in the training dataset and 0.899 in the test dataset, indicating excellent discrimination performance. In contrast, the AUC values for the risk prediction model constructed by logistic regression were 0.885 in the training dataset and 0.814 in the test dataset. The differences in AUC values between the random forest algorithm prediction model and the logistic regression risk prediction model highlight the superior discriminative performance of the random forest algorithm prediction model. The importance of the 11 variables in the Random Forest categorical prediction model of this study is detailed in Table 4 and Fig. 10 . The results of the DVT group columns, indicating the importance of the variables, highlight fracture onset to surgery time with the highest value at 22.611865, signifying its paramount importance in this DVT prediction model. In contrast, cardiopulmonary diseases have the lowest value of 6.535452, suggesting relatively lesser importance for DVT prediction among these 11 variables. The mean decrease accuracy column reveals that excluding the fracture onset to surgery time variable results in a significant 27.93% reduction in model accuracy, emphasizing its key role in accuracy. On the other hand, the mean reduced Gini coefficient indicates that higher values of the variable are more effective at segmenting the data. In this predictive model, fracture onset to surgery time emerges as the most crucial variable, playing a pivotal role in accurately predicting the occurrence of DVT. Additionally, D-dimer, PLT, and ALB are also important factors in predicting DVT occurrence. Overall, this Random Forest model heavily relies on ALB, PLT, and fracture onset to surgery time for accurate predictions. The validation results reveal that the method demonstrates strong discriminative properties, achieving impeccable classification across various metrics. In the training set, the 95% Confidence Interval (CI), Sensitivity, Specificity, Precision, Accuracy, and Balanced Accuracy stand at (0.9883, 1), 1.000, 1.000, 1.000, 1.000, and 1.000, respectively. For the test set, the 95% CI, Sensitivity, Specificity, Precision, Accuracy, and Balanced Accuracy are (0.7875, 0.9124), 0.6061, 0.9406, 0.7692, 0.8582, and 0.7733, respectively. This comprehensive evaluation underscores the method's robust performance and its ability to maintain high accuracy and balanced metrics in both training and test scenarios. The reliability of the predictive model was assessed using the Kappa statistic and F1 Score. In the training dataset, the Kappa values for the random forest-based prediction model and the logistic regression-based prediction model were 1.000 and 0.4815, respectively. In the test dataset, these values were 0.5887 and 0.3761, respectively. Examining the F1 Score, the random forest-based prediction model achieved a score of 1.000 in the training dataset, while the logistic regression-based prediction model scored 0.6408. In the test dataset, the F1 scores were 0.6780 for the random forest model and 0.5682 for the logistic regression model. These metrics provide insights into the model's reliability, with the random forest model consistently outperforming the logistic regression model across both training and test datasets. Table 4 Importance, MeanDecrease Accuracy and MeanDecrease Gini (important variables). Variables No-DVT group Importance DVT group Importance MeanDecrease Accuracy MeanDecrease Gini Cardiopulmonary 7.761557 6.535452 9.311122 2.524233 Fracture onset to surgery time 19.131834 27.935109 27.935109 20.649274 Fib 6.805638 8.437195 9.815045 8.929377 FDP 4.251672 6.606914 7.641734 6.653149 D-dimer 12.040965 11.216333 15.779695 12.408828 PLT 6.325702 13.072805 13.072691 11.943518 ALB 7.765926 20.678917 18.478731 14.203333 A/G 6.176767 6.718918 8.463501 11.171071 CREA 6.496627 8.778142 10.283946 8.467382 GFR 4.126512 6.750375 7.565805 9.319140 Pi 6.481033 7.447505 9.497667 9.349841 Discussion With the aging of the world's population, there is a growing interest in managing elderly HF patients[ 21 ]. Factors such as age, gender, low serum vitamin D levels, and low bone mineral density are significantly associated with osteoporosis[ 21 ]. The elderly population is more susceptible to HFs. DVT often develops in elderly patients suffering HFs due to limited mobility, vascular injury, altered coagulation factors post-fracture, surgery, and diabetic comorbidities[ 12 , 13 , 22 ]. DVT is more perilous than other HF complications as it can lead to life-threatening diseases like pulmonary embolism, posing a threat to the lives of elderly patients[ 23 ]. However, diagnosing DVT in elderly patients with fractures is limited by the prevalence of comorbidities that may mask symptoms and complicate accurate assessments. Predicting DVT in elderly HF patients soon after hospital admission is crucial for early intervention, personalized treatment plans, and reducing the risk of postoperative mortality[ 22 ]. It also enhances clinical prognosis. While current DVT prediction models often amalgamate risk factors using methods like univariate regression followed by multivariate logistic regression, they exhibit limitations in accurately predicting outcomes for individual patients in complex elderly cases. In response, we developed a robust random forest prediction model, demonstrating significant discriminatory power for predicting deep vein thrombosis. The ROC visualization analysis underscored the superior predictive accuracy of the random forest algorithm over classical logistic regression in both the training and test datasets. Random forest prediction models are ensemble learning algorithms that construct a multitude of decision trees during the training phase and subsequently leverage these trees for making predictions[ 24 ]. Each decision tree is trained on a distinct subset of the data, and the model consolidates the individual predictions to arrive at a final decision. The unique strength of the Random Forest model lies in its capacity to manage large datasets with high dimensionality, even encompassing thousands of clinical variables[ 25 ]. It achieves this by mitigating overfitting through the averaging of predictions from multiple trees and by providing rankings of feature importance. The advantages of random forest models over other models include robustness to noisy data, the ability to handle both classification and regression tasks, and resistance to overfitting, ultimately leading to enhanced generalization performance[ 26 ].Our study marks the first attempt to integrate the Random Forest algorithm into predicting admission thrombosis in elderly hip fracture (HF) patients. This integration aims to develop a prediction model using basic characteristics of admitted HF patients, their medical history, and essential preoperative laboratory blood test indicators. The Random Forest prediction model has demonstrated significant potential for foreseeing thrombosis in newly admitted elderly HF patients. As machine learning, particularly the Random Forest algorithm, continues to evolve and validate, coupled with the abundance of clinical data, the exploration and development of Random Forest predictive models for various diseases warrant considerable attention[ 25 ].Compared to traditional logistic regression, random forests are less sensitive to multicollinearity and exhibit relative robustness to missing data[ 27 ]. The predictive results of the random forest algorithm remain stable and are less influenced by existing missing data compared to logistic regression. With the advancement of big data, the prevalence of multicenter clinical studies continues to rise. In scenarios involving several thousand variables, logistic regression proves inadequate, while the random forest algorithm can offer more accurate predictions. In this study, we found that 25% of elderly hip fracture (HF) patients who underwent surgical intervention during hospitalization developed deep vein thrombosis (DVT). The prediction model focused on easily obtainable clinical variables to enable swift and efficient prediction. The Random Forest algorithm highlighted specific variables, such as ALB (albumin), A/G (albumin-to-globulin ratio), PLT (platelet count), Fib (fibrinogen), D-dimer, CREA (creatinine), Pi (inorganic phosphorus), GFR (estimated glomerular filtration rate), FDP (fibrin and fibrinogen degradation products), fracture onset to surgery time, and cardiopulmonary diseases, as valuable for characterizing and predicting preoperative thrombosis in elderly Chinese HF patients during hospitalization. Subsequently, we validated these nine variables using Lasso regression modeling. In previous studies, age at surgery has been recognized as a significant risk factor for preoperative DVT in HF patients, particularly among the elderly compared to other age groups. In Wang et al.'s study, surgical age was classified into four groups: 60–70 years, 70–80 years, 80–90 years, and > 90 years[ 9 ]. The results indicated a heightened risk of preoperative DVT in patients over 90 years, with no statistically significant differences among the other age groups[ 9 ]. This observation may explain the lack of significance of age as a variable in the Random Forest predictive model used in this study. The study focused on five laboratory test indicator variables, which have been previously documented in the literature. In the investigation exploring the correlation between cardiovascular diseases and the formation of deep vein thrombosis (DVT), several significant studies have been conducted. Gregson et al.[ 11 ] carried out a prospective cohort study to examine the association between cardiovascular risk factors and venous thromboembolism (VTE), which includes DVT and pulmonary embolism (PE). They identified older age, smoking, and obesity as overlapping risk factors for both conditions. Liu et al.'s multicenter cohort study revealed a 24.5% incidence of venous thromboembolism (VTE) in patients with acute exacerbation of chronic obstructive pulmonary disease (AECOPD) [ 28 ]. The major risk factors identified included older age, higher body mass index, and a history of VTE. An interesting finding emerged regarding the impact of DVT formation on acute cardiopulmonary symptoms, including asymptomatic pulmonary embolism. Hou et al.[ 29 ] investigated the association between DVT and pulmonary embolism (PE). Their results indicated that the occurrence of PE in patients with lower extremity deep vein thrombosis is associated with alcohol consumption and heart failure. The proximal veins were more affected than the distal ones, and the right side was more affected than the left. Numerous studies have demonstrated the biological mechanisms by which cardiovascular diseases lead to deep vein thrombosis (DVT)[ 30 – 32 ]: Hemodynamic changes: On one hand, conditions such as heart failure can result in reduced cardiac output, thereby causing blood flow to slow down. This stasis can promote the formation of deep vein thrombosis. On the other hand, prolonged inactivity or bed rest, which is common in patients with hip fractures, can lead to venous stasis. This increases the risk of DVT as blood accumulates in the veins. Endothelial dysfunction: On one hand, cardiovascular diseases such as atherosclerosis can damage the inner layer of blood vessels. This damage exposes the underlying tissue, activates the coagulation cascade, and promotes thrombosis. On the other hand, chronic inflammation associated with cardiovascular diseases can impair endothelial function. Inflammatory markers such as C-reactive protein (CRP) and interleukin-6 (IL-6) can promote a pre-thrombotic state by upregulating adhesion molecules and procoagulant factors on endothelial cells. Hypercoagulable state: Cardiovascular diseases can lead to an increase in coagulation factors, such as fibrinogen, factor VII, and factor VIII. These factors enhance the coagulation process and increase the risk of thrombosis. Additionally, obesity, which is a risk factor for cardiovascular diseases, can cause an increase in the levels of plasminogen activator inhibitor-1 (PAI-1), inhibiting the conversion of plasminogen to plasmin and thus reducing the breakdown of fibrin clots, leading to a decrease in the activity of the fibrinolytic system. The impact of the time from fracture occurrence to surgical intervention on the incidence of deep vein thrombosis (DVT) has been extensively studied. Yang et al.'s research emphasizes the critical role of delayed time from injury to surgery in increasing the risk of preoperative deep vein thrombosis in fracture patients[ 33 ]. This retrospective study underscores the importance of timely surgical intervention, particularly in elderly patients, for preventing DVT. In a study focusing on hip fractures published in the "Journal of Orthopaedic Surgery and Research," Taoka et al.[ 34 ] found that a delay of more than 48 hours in surgical intervention after hip fracture significantly increases the risk of developing proximal DVT. Prolonged bed rest leads to slow blood flow in the lower limbs, especially in the deep veins. Blood stasis provides an opportunity for thrombus formation. At the same time, due to hip fractures, there is a lack of muscle activity in the legs, which normally helps to propel blood upward into the veins, leading to blood pooling. The longer the delay in surgery, the more pronounced these hemodynamic changes become, thereby increasing the risk of DVT. When a blood clot forms in the body, the coagulation system is activated, leading to the conversion of fibrinogen to fibrin. The fibrin forms a mesh-like structure to stabilize the clot. Subsequently, the fibrinolytic system is also activated to break down the fibrin clot. D-dimer is a specific degradation product of cross-linked fibrin. Its presence and increased levels in the blood indicate that there has been recent fibrin formation and subsequent degradation, which is a key process in thrombosis. Elevated D-dimer levels, coupled with indications of an active fibrinolytic system, suggest the potential presence of a clot, thereby increasing the risk of thrombotic events such as DVT in elderly patients with hip fractures (HF)[ 6 ]. Some studies have noted age-dependent variations in the specificity of D-dimer, with reduced accuracy, particularly in individuals over 65 years old. In a retrospective study, D-dimer and fibrinogen levels emerged as potential predictors of postoperative thrombosis in patients with lower extremity fractures[ 35 ]. Beyond D-dimer, other coagulation function indicators, such as Fib and FDP, have also emerged as potential biomarkers for thrombus formation[ 36 ]. For instance, Bai et al.[ 37 ] investigated thrombotic risk prediction in patients with acute promyelocytic leukemia (APL) and observed significantly elevated levels of FDP and D-dimer in APL patients with thrombosis. The researchers proposed that the ratios of FDP/Fib and D-dimer/Fib could be significantly correlated factors in predicting thrombus formation. These ratios also hold value as prognostic indicators, particularly in high-risk patients prone to thrombosis. Similarly, Pang et al.[ 38 ] identified elevated levels of D-dimer and fibrinogen degradation products in patients with post-mastectomy DVT, emphasizing their importance as valuable indicators of thrombus formation. Additionally, Yang et al.'s research indicated that FDP is an independent risk factor and a significant predictor of deep vein thrombosis in patients with acute coronary syndrome, underscoring its importance as a valuable marker in thrombus formation[ 39 ]. Albumin (ALB), a major plasma protein, plays a crucial role in maintaining vascular integrity and preventing thrombosis. Studies have shown[ 40 ] that ALB can bind with arachidonic acid, preventing its metabolism into potent aggregating substances, endoperoxides, and thromboxane A2, thereby inhibiting platelet activation and aggregation. In addition, it exhibits a concentration-dependent ability to induce inducible nitric oxide synthase in macrophages, thereby increasing the production of nitric oxide, a potent platelet inhibitor. In the presence of inflammation, increased exertion, or injury, ALB levels typically decrease, making it a potential predictor of deep vein thrombosis (DVT) in elderly patients with hip fractures (HFs). Despite ALB's dependence on various physiological factors and health status, caution is warranted in interpreting it as an independent predictor. However, the random forest prediction model employed in this study allows for analytical prediction by integrating diverse variables. Wu et al.[ 41 ] conducted a retrospective examination of the correlation between serum albumin levels and preoperative DVT in patients aged 65 years and older with HF. The results indicated a 6% decrease in DVT risk for every 1 g/L increase in albumin concentration after accounting for confounding factors, suggesting its potential as a predictor of DVT risk in the elderly. Yao et al.[ 42 ] utilized the D-dimer to albumin ratio (DAR) to predict perioperative DVT in elderly patients with HF, achieving a final AUC of 0.677. Platelets are one of the main participants in the processes of thrombosis formation and hemostasis[ 43 ]. Elevated platelet levels have the potential to initiate the coagulation cascade and enhance platelet self-aggregation, making them potential predictors of deep vein thrombosis (DVT). In the Nomogram prediction model for preoperative DVT in the elderly developed by Zhang et al.[ 44 ], the platelet (PLT) count emerged as one of the independent risk factors, showing a statistical difference between groups. Similarly, a PLT count of 220 × 10^9/L was identified as an independent factor for DVT in a study by Niu et al.[ 45 ]. The risk of DVT in elderly hip fracture (HF) patients with preoperative levels exceeding 220 × 10^9/L was 2.02 times higher than in patients without DVT. Wang et al.[ 46 ] demonstrated a significant correlation, indicating that a mean platelet volume (MPV) level < 13.3 fL was associated with DVT. Creatinine (CREA) serves as an indicator of renal function, and its elevation typically arises due to impaired renal function. Studies have shown[ 47 ] that elevated creatinine levels are associated with increased levels of coagulation factors (such as FVIII), which can lead to a hypercoagulable state in the blood and increase the risk of thrombosis. When kidney function is impaired, changes in hemodynamics may affect the pressure of blood return in the lower limbs, leading to stasis of blood in the lower limb veins. On the other hand, the levels of metabolic waste and toxins in the blood (such as urea and creatinine) will increase, which may lead to an increase in blood viscosity. A creatinine level exceeding 0.96 mg/dL has been identified as a significant predictor of thromboembolic events, as demonstrated in a multifactorial analysis within a nested study conducted by Haltout et al.[ 48 ]. In a separate study, Ding et al. developed an XGBoost model for machine learning screening and predicting risk factors associated with deep vein thrombosis (DVT) following hip arthroplasty. Notably, CREA was integrated as a primary characterization indicator in this predictive model. The presence of chronic kidney disease (CKD) contributes to a decline in glomerular filtration rate (GFR), potentially inducing hemostatic changes that could influence the risk of DVT formation. In the groundbreaking study by Wang et al.[ 49 ], the first exploration of the relationship between chronic kidney dysfunction and preoperative DVT in hip fracture patients was conducted. Among patients with an estimated GFR (eGFR) less than 60 mL/min/1.73 m², the incidence of DVT was 6.0%. The study establishes that eGFR serves as an independent predictive factor for preoperative DVT formation. Additionally, elevated fibrinogen levels were identified as another independent predictive factor for DVT formation. Li et al.'s research revealed a 2.68-fold increase in the risk of DVT in stage 3/4 CKD patients[ 50 ]. Severe chronic kidney disease may render patients more susceptible to proximal and symptomatic deep vein thrombosis. CKD emerges as a significant risk factor for DVT occurrence after total hip arthroplasty. Organic phosphorus is involved in glycolysis, ammonification, and oxidative phosphorylation, generating chemical energy by producing adenosine triphosphate (ATP) from adenosine diphosphate (ADP). It also affects the oxygen-carrying capacity of hemoglobin by regulating the synthesis of 2,3-diphosphoglycerate. Phosphorus atoms are components of DNA and RNA bases, as well as phospholipids, which are involved in cell structure and signal transduction[ 51 ]. Therefore, a decrease in blood phosphorus levels in the human body can lead to cell damage, tissue hypoxia, and multi-organ damage, especially in vascular endothelial cells, causing the aggregation of platelets and coagulation factors. Elevated blood phosphorus levels are also an important factor in the occurrence of deep vein thrombosis (DVT)[ 52 ]. When blood phosphorus levels increase, beta-glycerophosphate in human endothelial cells induces endothelial cell damage by increasing the expression of p16 and the activity of age-related beta-galactosidase. Lu et al.[ 53 ] used serum phosphorus as a continuous variable. They identified risk factors for DVT following hip arthroplasty through a comprehensive analysis involving multivariate binary logistic regression and generalized additive models. Their research findings support the view that serum phosphorus levels can serve as a predictive indicator of the risk of developing DVT. Our study has several limitations. Firstly, it is retrospective in nature, which may introduce selection bias and limit the ability to establish causality. Secondly, our focus was solely on the incidence of lower extremity deep vein thrombosis (DVT), excluding other sites, which may not provide a comprehensive view of the overall thrombotic risk. Thirdly, our reliance on the quality of medical record data hindered patient validation, potentially affecting the accuracy of the results. Lastly, our study population consisted exclusively of elderly hip fracture (HF) patients undergoing surgical intervention during hospitalization, rather than encompassing all elderly HF patients, which resulted in an insufficiently large sample size. To enhance the accuracy of the random forest prediction model, future research should involve extensive multicenter samples to increase the sample size and improve the generalizability of the findings. Additionally, a prospective study design could help mitigate some of the limitations associated with retrospective data collection. Including a broader range of patients, such as those with different types of fractures or those managed conservatively, would provide a more comprehensive understanding of the risk factors for DVT. Furthermore, incorporating more detailed and accurate data collection methods, possibly through direct patient interviews or more rigorous data entry protocols, could enhance the quality of the data and the robustness of the model. Conclusions In this study, we successfully developed and validated a Random Forest-based predictive model to anticipate preoperative deep vein thrombosis (DVT) in elderly hip fracture (HF) patients. The model demonstrated robust discriminative performance, achieving an AUC of 1.000 in the training set and 0.899 in the test set, significantly outperforming traditional logistic regression approaches. Key preoperative indicators—including ALB, D-dimer, PLT, fracture-to-surgery time, and cardiopulmonary comorbidities—were identified as critical predictors, aligning with pathophysiological mechanisms of thrombosis. This model not only enhances preoperative risk stratification but also provides clinicians with actionable insights to implement timely interventions, such as early anticoagulation or optimized surgical scheduling, thereby reducing perioperative morbidity and mortality. Future research will focus on expanding the sample size and refining the model to further improve its predictive accuracy and clinical utility. Declarations Author Contributions ： Xuehai Jia and Hao Liu were responsible for data analysis and drafting the manuscript. Kerui Zhang provided technical guidance. Yi Deng, Changyong Shen and Ya Li were responsible for data preprocessing. Litai Ma and Fei Xing were responsible for the experimental design and paper review. All authors read and approved the final manuscript. Conflict of Interest declaration : The authors declare that they have no affiliations with or involvement in any organization or entity with any financial interest in the subject matter or materials discussed in this manuscript. Ethics Statement ： The study was conducted in accordance with the Declaration of Helsinki. The data of human received ethical approval from the Ethics Review Committee of West China Hospital of Sichuan University (Approval Number: 2023 Review No. 2427). Informed consent was obtained from all subjects whose CT images were used in this study, and all data were anonymized to protect patient p rivacy. Funding Information ： The present study received no financial support from either public, commercial, or not-for-profit sources. Consent for publication: Not applicable. Availability of data and materials: The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Acknowledgements: The authors thank the orthopedists, Fuguo Huang, Yue Fang and Shiqiang Cen, for their help in this study. References S. Strøm Rönnquist, B. Viberg, M.T. Kristensen, H. Palm, J.-E.B. Jensen, C.F. Madsen, K.E. Åkesson, S. Overgaard, C. Rogmark, Frailty and osteoporosis in patients with hip fractures under the age of 60—a prospective cohort of 218 individuals, Osteoporosis International 33(5) (2022) 1037-1055. Y.W. Zhang, P.P. Lu, Y.J. Li, G.C. Dai, M.H. Chen, Y.K. Zhao, M.M. Cao, Y.F. Rui, Prevalence, Characteristics, and Associated Risk Factors of the Elderly with Hip Fractures: A Cross-Sectional Analysis of NHANES 2005-2010, Clin Interv Aging 16 (2021) 177-185. Y.W. Zhang, M.M. Cao, Y.J. Li, G.C. Dai, P.P. Lu, M. Zhang, L.Y. Bai, X.X. Chen, L. Shi, C. Zhang, Y.F. Rui, Dietary Protein Intake in Relation to the Risk of Osteoporosis in Middle-Aged and Older Individuals: A Cross-Sectional Study, J Nutr Health Aging 26(3) (2022) 252-258. Y.W. Zhang, P.P. Lu, Y.J. Li, G.C. Dai, M.M. Cao, T. Xie, C. Zhang, L. Shi, Y.F. Rui, Low dietary choline intake is associated with the risk of osteoporosis in elderly individuals: a population-based study, Food Funct 12(14) (2021) 6442-6451. B.L.L. Wong, Y.H. Chan, G.K. O'Neill, D. Murphy, R.A. Merchant, Frailty, length of stay and cost in hip fracture patients, Osteoporos Int 34(1) (2023) 59-68. L. Li, J. Zhen, L. Huang, J. Zhou, L. Yao, L. Xu, W. Zhang, G. Zhang, Q. Chen, B. Cheng, S. Gong, G. Cai, R. Jiang, J. Yan, The risk factors for deep venous thrombosis in critically ill older adult patients: a subgroup analysis of a prospective, multicenter, observational study, BMC Geriatrics 22(1) (2022) 977. V. Sathiyakumar, S.E. Greenberg, A.A. Jahangir, H.H. Mir, W.T. Obremskey, M.K. Sethi, Impact of type of surgery on deep venous thrombi and pulmonary emboli: a look at twenty seven thousand hip fracture patients, Int Orthop 39(10) (2015) 2017-22. Y.C. Lin, S.H. Lee, I.J. Chen, C.H. Chang, C.J. Chang, Y.C. Wang, Y. Chang, P.H. Hsieh, Symptomatic pulmonary embolism following hip fracture: A nationwide study, Thromb Res 172 (2018) 120-127. T. Wang, J. Guo, Y. Long, Y. Yin, Z. Hou, Risk factors for preoperative deep venous thrombosis in hip fracture patients: a meta-analysis, J Orthop Traumatol 23(1) (2022) 19. D. Kumar, A. Elhence, R.K. Rajnish, N. Gahlot, S.K. Yadav, S. Gupta, K. Kunal, P. Khera, Mechanical thromboprophylaxis for hip fractures in elderly patients: a prospective randomized controlled study, Am J Transl Res 15(8) (2023) 5284-5291. J. Gregson, S. Kaptoge, T. Bolton, L. Pennells, P. Willeit, S. Burgess, S. Bell, M. Sweeting, E.B. Rimm, C. Kabrhel, B. Zöller, G. Assmann, V. Gudnason, A.R. Folsom, V. Arndt, A. Fletcher, P.E. Norman, B.G. Nordestgaard, A. Kitamura, B.K. Mahmoodi, P.H. Whincup, M. Knuiman, V. Salomaa, C. Meisinger, W. Koenig, M. Kavousi, H. Völzke, J.A. Cooper, T. Ninomiya, E. Casiglia, B. Rodriguez, Y. Ben-Shlomo, J.-P. Després, L. Simons, E. Barrett-Connor, C. Björkelund, M. Notdurfter, D. Kromhout, J. Price, S.E. Sutherland, J. Sundström, J. Kauhanen, J. Gallacher, J.W.J. Beulens, R. Dankner, C. Cooper, S. Giampaoli, J.F. Deen, A. Gómez de la Cámara, L.H. Kuller, A. Rosengren, P.J. Svensson, D. Nagel, C.J. Crespo, H. Brenner, J.R. Albertorio-Diaz, R. Atkins, E.J. Brunner, M. Shipley, I. Njølstad, D.A. Lawlor, Y.T. van der Schouw, R.M. Selmer, M. Trevisan, W.M.M. Verschuren, P. Greenland, S. Wassertheil-Smoller, G.D.O. Lowe, A.M. Wood, A.S. Butterworth, S.G. Thompson, J. Danesh, E. Di Angelantonio, T. Meade, f.t.E.R.F. Collaboration, Cardiovascular Risk Factors Associated With Venous Thromboembolism, JAMA Cardiology 4(2) (2019) 163-173. M. Cushman, Epidemiology and risk factors for venous thrombosis, Semin Hematol 44(2) (2007) 62-9. F.A. Anderson, F.A. Spencer, Risk Factors for Venous Thromboembolism, Circulation 107(23_suppl_1) (2003) I-9-I-16. B.-H. Wang, Y.-D. Sun, X.-C. Fan, B.-F. Zhang, The admission pH is a risk factor of preoperative deep vein thrombosis in geriatric hip fracture: a retrospective cohort study, Scientific Reports 13(1) (2023) 18392. J. Jiang, F. Xing, R. Luo, Z. Chen, H. Liu, Z. Xiang, X. Duan, Risk factors and prediction model of nomogram for preoperative calf muscle vein thrombosis in geriatric hip fracture patients, Frontiers in Medicine 10 (2023). L. Langsetmo, J.T. Schousboe, B.C. Taylor, J.A. Cauley, H.A. Fink, P.M. Cawthon, D.M. Kado, K.E. Ensrud, Advantages and Disadvantages of Random Forest Models for Prediction of Hip Fracture Risk Versus Mortality Risk in the Oldest Old, JBMR Plus 7(8) (2023) e10757. J. Wu, J. Qiu, E. Xie, W. Jiang, R. Zhao, J. Qiu, M.A. Zafar, Y. Huang, C. Yu, Predicting in-hospital rupture of type A aortic dissection using Random Forest, Journal of Thoracic Disease 11(11) (2019) 4634-4646. X. Han, X. Zheng, Y. Wang, X. Sun, Y. Xiao, Y. Tang, W. Qin, Random forest can accurately predict the development of end-stage renal disease in immunoglobulin a nephropathy patients, Ann Transl Med 7(11) (2019) 234. P.C. Kruger, J.W. Eikelboom, J.D. Douketis, G.J. Hankey, Deep vein thrombosis: update on diagnosis and management, Med J Aust 210(11) (2019) 516-524. L. Duffett, Deep Venous Thrombosis, Annals of Internal Medicine 175(9) (2022) ITC129-ITC144. L.-K. Liu, W.-J. Lee, L.-Y. Chen, A.-C. Hwang, M.-H. Lin, L.-N. Peng, L.-K. Chen, Association between Frailty, Osteoporosis, Falls and Hip Fractures among Community-Dwelling People Aged 50 Years and Older in Taiwan: Results from I-Lan Longitudinal Aging Study, PLOS ONE 10(9) (2015) e0136968. J. Stone, P. Hangge, H. Albadawi, A. Wallace, F. Shamoun, M.G. Knuttien, S. Naidu, R. Oklu, Deep vein thrombosis: pathogenesis, diagnosis, and medical management, Cardiovascular Diagnosis and Therapy (2017) S276-S284. Y. Zhou, H. Chen, Y. Zhang, C. Yang, X. Yi, S. Liu, Y. Zeng, Computational Analysis of the Related Factors of Deep Vein Thrombosis (DVT) Formation in Patients Undergoing Hip Fracture Surgery, Evidence-Based Complementary and Alternative Medicine 2022 (2022) 1127095. E. Kavalci, A. Hartshorn, Improving clinical trial design using interpretable machine learning based prediction of early trial termination, Sci Rep 13(1) (2023) 121. M. Toma, O.C. Wei, Predictive Modeling in Medicine, Encyclopedia 3(2) (2023) 590-601. K.-M. Kim, J.-H. Kim, H.-S. Rhee, B.-Y. Youn, Development of a prediction model for the depression level of the elderly in low-income households: using decision trees, logistic regression, neural networks, and random forest, Scientific Reports 13(1) (2023) 11473. F. Xing, R. Luo, M. Liu, Z. Zhou, Z. Xiang, X. Duan, A New Random Forest Algorithm-Based Prediction Model of Post-operative Mortality in Geriatric Patients With Hip Fractures, Frontiers in Medicine 9 (2022). X. Liu, X. Jiao, X. Gong, Q. Nie, Y. Li, G. Zhen, M. Cheng, J. He, Y. Yuan, Y. Yang, Prevalence, Risk Factor and Clinical Characteristics of Venous Thrombus Embolism in Patients with Acute Exacerbation of COPD: A Prospective Multicenter Study, Int J Chron Obstruct Pulmon Dis 18 (2023) 907-917. J. Hou, W. Wang, H. Cai, J. Chen, B. Chen, Z. Shen, Y. Tang, J. Li, S. Liu, Y. Mei, J. Wang, S. Lu, Patients With Right Lower Extremity Deep Vein Thrombosis Have a Higher Risk of Symptomatic Pulmonary Embolism: A Retrospective Study of 1585 Patients, Annals of Vascular Surgery 81 (2022) 240-248. J.R. Byrnes, A.S. Wolberg, Red blood cells in thrombosis, Blood 130(16) (2017) 1795-1799. T. Watson, E. Shantsila, G.Y. Lip, Mechanisms of thrombogenesis in atrial fibrillation: Virchow's triad revisited, Lancet 373(9658) (2009) 155-66. G. Alkarithi, C. Duval, Y. Shi, F.L. Macrae, R.A.S. Ariens, Thrombus Structural Composition in Cardiovascular Disease, Arterioscler Thromb Vasc Biol 41(9) (2021) 2370-2383. W. Yang, Q. Wei, H. Wang, K. Ding, M. Li, C. Li, C. Liang, Y. Zhu, W. Chen, Preoperative incidence and risk factors of deep venous thrombosis in patients with isolated femoral shaft fracture, BMC Surgery 22(1) (2022) 83. T. Taoka, T. Ohmori, T. Kanazawa, K. Toda, T. Ishihara, Y. Ito, Delayed surgery after hip fracture affects the incidence of venous thromboembolism, Journal of Orthopaedic Surgery and Research 18(1) (2023) 630. J. Cheng, Z.b. Fu, J. Zhu, L. Zhou, W. Song, The predictive value of plasminogen activator inhibitor-1, fibrinogen, and D-dimer for deep venous thrombosis following surgery for traumatic lower limb fracture, Annals of palliative medicine (2020). T. Wang, J. Guo, Y. Long, Y. Yin, Z. Hou, Risk factors for preoperative deep venous thrombosis in hip fracture patients: a meta-analysis, Journal of Orthopaedics and Traumatology 23(1) (2022) 19. Y. Bai, M. Shi, X. Yang, W. Zhang, R. Yang, X. Wei, X. Wei, L. Duan, C. Wang, R. Mi, H.A. Waqas Ahmed, L. Huo, Y. Chen, F. Xu, D. Wu, K. Sun, The value of FDP/FIB and D-dimer/FIB ratios in predicting high-risk APL-related thrombosis, Leukemia Research 79 (2019) 34-37. M. Pang, F. Zhao, P. Yu, X. Zhang, H. Xiao, W. Qiang, H. Zhu, L. Zhao, The significance of coagulation and fibrinolysis-related parameters in predicting postoperative venous thrombosis in patients with breast cancer, Gland Surg 10(4) (2021) 1439-1446. S. Yang, T. Wang, Y. Long, J. Guo, Z. Hou, Risk Factors of Deep Vein Thrombosis in Adults with Acute Compartment Syndrome Following Lower Extremity Fractures, Clinical and Applied Thrombosis/Hemostasis 29 (2023) 10760296231165053. Y. Sun, J. Deng, Y. Ding, S. Luo, S. Li, Y. Guan, X. Cao, X. Hao, Y. Hu, Serum albumin, genetic susceptibility, and risk of venous thromboembolism, Research and Practice in Thrombosis and Haemostasis 8(5) (2024). Y.-L. Wu, D. Zhang, K.-Y. Zhang, T. Yan, W.-S. Qiang, T. Zhang, B.-F. Zhang, The association between admission serum albumin and preoperative deep venous thrombosis in geriatrics hip fracture: a retrospective study of 1819 patients with age ≥ 65 years, BMC Musculoskeletal Disorders 24(1) (2023) 672. W. Yao, K. Zhang, Q. Lv, Z. Deng, W. Ding, D-dimer-albumin ratio (DAR) as a new biomarker for predicting preoperative deep vein thrombosis after geriatric hip fracture patients, Journal of Orthopaedic Surgery and Research 18(1) (2023) 645. N. Cay, A. Ipek, M. Gumus, Z. Birkan, E. Ozmen, Platelet activity indices in patients with deep vein thrombosis, Clin Appl Thromb Hemost 18(2) (2012) 206-10. L. Zhang, M. He, W. Jia, W. Xie, Y. Song, H. Wang, J. Peng, Y. Li, Z. Wang, Z. Lin, Analysis of high-risk factors for preoperative DVT in elderly patients with simple hip fractures and construction of a nomogram prediction model, BMC Musculoskeletal Disorders 23(1) (2022) 441. S. Niu, J. Li, Y. Zhao, D. Ding, G. Jiang, Z. Song, Preoperative deep venous thrombosis (DVT) after femoral neck fracture in the elderly, the incidence, timing, location and related risk factors, BMC Musculoskeletal Disorders 22(1) (2021) 264. Z. Wang, X. Chen, J. Wu, Q. Zhou, H. Liu, Y. Wu, S. Liu, Y. Liu, Low Mean Platelet Volume is Associated with Deep Vein Thrombosis in Older Patients with Hip Fracture, Clinical and Applied Thrombosis/Hemostasis 28 (2022) 10760296221078837. M.J. Huang, R.B. Wei, Y. Wang, T.Y. Su, P. Di, Q.P. Li, X. Yang, P. Li, X.M. Chen, Blood coagulation system in patients with chronic kidney disease: a prospective observational study, BMJ Open 7(5) (2017) e014294. J. Haltout, A. Awada, M. Paesmans, M. Moreau, J. Klastersky, G. Machiels, M. Ignatiadis, N. Kotecki, Predictive factors for cancer-associated thrombosis in a large retrospective single-center study, Supportive Care in Cancer 27(4) (2019) 1163-1170. Z. Wang, J. Xiao, Z. Zhang, X. Qiu, Y. Chen, Chronic kidney disease can increase the risk of preoperative deep vein thrombosis in middle-aged and elderly patients with hip fractures, Clin Interv Aging 13 (2018) 1669-1674. Q. Li, B. Dai, Y. Yao, K. Song, D. Chen, Q. Jiang, Chronic Kidney Dysfunction Can Increase the Risk of Deep Vein Thrombosis after Total Hip and Knee Arthroplasty, BioMed Research International 2017 (2017) 8260487. D. Gruson, A. Buglioni, J.C. Burnett, Jr., PTH: Potential role in management of heart failure, Clin Chim Acta 433 (2014) 290-6. G. Olmos, P. Martinez-Miguel, E. Alcalde-Estevez, D. Medrano, P. Sosa, L. Rodriguez-Manas, M. Naves-Diaz, D. Rodriguez-Puyol, M.P. Ruiz-Torres, S. Lopez-Ongil, Hyperphosphatemia induces senescence in human endothelial cells by increasing endothelin-1 production, Aging Cell 16(6) (2017) 1300-1312. D.-X. Lu, K. Zhang, T. Ma, M. Li, Z. Li, Y.-B. Xu, C.-F. Wang, C. Ren, B.-F. Zhang, The Association between Admission Serum Phosphorus and Preoperative Deep Venous Thrombosis in Geriatric Hip Fracture: A Retrospective Study, Diagnostics 13(3) (2023) 545. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6339181","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":490383653,"identity":"a125e0a7-f8bb-4a62-90e9-4683e15dae41","order_by":0,"name":"Xuehai Jia","email":"","orcid":"","institution":"West China Hospital of Sichuan University","correspondingAuthor":false,"prefix":"","firstName":"Xuehai","middleName":"","lastName":"Jia","suffix":""},{"id":490383654,"identity":"65232292-45d3-4a30-be5c-b6fa5371ffaf","order_by":1,"name":"Hao Liu","email":"","orcid":"","institution":"West China Hospital of Sichuan University","correspondingAuthor":false,"prefix":"","firstName":"Hao","middleName":"","lastName":"Liu","suffix":""},{"id":490383655,"identity":"f280a1fb-a82a-4eae-a35b-452597b960da","order_by":2,"name":"Changyong Shen","email":"","orcid":"","institution":"Department of Orthopedics, Yibin Third People's Hospital, Yibin, Sichuan, China","correspondingAuthor":false,"prefix":"","firstName":"Changyong","middleName":"","lastName":"Shen","suffix":""},{"id":490383656,"identity":"1c4f5ace-0639-471a-8d79-8f5d031b80bd","order_by":3,"name":"Yi Deng","email":"","orcid":"","institution":"West China Hospital of Sichuan University","correspondingAuthor":false,"prefix":"","firstName":"Yi","middleName":"","lastName":"Deng","suffix":""},{"id":490383657,"identity":"6cdefb5e-b2c0-47f6-a2cb-fa635d4d7b98","order_by":4,"name":"Kerui Zhang","email":"","orcid":"","institution":"West China Hospital of Sichuan University","correspondingAuthor":false,"prefix":"","firstName":"Kerui","middleName":"","lastName":"Zhang","suffix":""},{"id":490383658,"identity":"35412589-ec72-4ebc-95ef-9d75c55b314c","order_by":5,"name":"Litai Ma","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAtklEQVRIie3ROwrCQBCA4QkLVqNpI1h4AWEhkCroQWwGAuk8gQEDFim90siANkJay80NLFO6SWW3U1rsX8/H7AMgFvvPEiYATI0RpzZMBLjuFrXVE/BrbI/bTDW+6xh4GPebXBAsNOUxSIoXTQersJAlO3jUpzZIeCbGkxXZpBUF6d1ELphf0WY68p63CFqjJ0PLVD8xE//IpLpLX92HsTwf0puI+zRlmPjP/5mh8HgsFovFNH0BGCo/RIoOb7YAAAAASUVORK5CYII=","orcid":"","institution":"West China Hospital of Sichuan University","correspondingAuthor":true,"prefix":"","firstName":"Litai","middleName":"","lastName":"Ma","suffix":""},{"id":490383659,"identity":"8cf38fb3-e982-4ce4-82c5-e6b29b201a28","order_by":6,"name":"Fei Xing","email":"","orcid":"","institution":"West China Hospital of Sichuan University","correspondingAuthor":false,"prefix":"","firstName":"Fei","middleName":"","lastName":"Xing","suffix":""}],"badges":[],"createdAt":"2025-03-30 15:08:15","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6339181/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6339181/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":87849715,"identity":"a09fa994-d48e-4465-98c1-cced9e87b7e6","added_by":"auto","created_at":"2025-07-29 15:46:25","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":164237,"visible":true,"origin":"","legend":"\u003cp\u003eThe steps involved in establishing, testing, and validating DVT Random Forest predictive model.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6339181/v1/80909ed93fe2dad2248f2175.png"},{"id":87849714,"identity":"5e90b499-9760-4003-95aa-bca6f4c22993","added_by":"auto","created_at":"2025-07-29 15:46:25","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":121638,"visible":true,"origin":"","legend":"\u003cp\u003eHistogram of the AUC composition of each variable in the univariate analysis.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6339181/v1/39322c3dbc401916643b7d9e.png"},{"id":87849717,"identity":"50273c88-5bd3-4665-982f-0e2595d61563","added_by":"auto","created_at":"2025-07-29 15:46:25","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":249866,"visible":true,"origin":"","legend":"\u003cp\u003eThe top 16 continuous variables were visualized and observed according to the AUC ranking in Figure 2. prothrombin time (PT), International Normalized Ratio (INR), fibrinogen (Fib), fibrin and fibrinogen degradation products (FDP), D-dimer, platelet count (PLT), white blood cell count (WBC), aspartate aminotransferase (AST), albumin (ALB), albumin-to-globulin ratio(A/G), creatinine (CREA), estimated glomerular filtration rate (GFR), triglyceride (TG), creatine kinase (CK), lactate dehydrogenase (LDH), and inorganic phosphorus (Pi).\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-6339181/v1/1ee074cc87349e3a948449ab.png"},{"id":87851198,"identity":"273c335d-a8c7-47b1-942c-0317d7b3b8aa","added_by":"auto","created_at":"2025-07-29 15:54:25","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":164060,"visible":true,"origin":"","legend":"\u003cp\u003eVisualization of AUC for dichotomous variables\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-6339181/v1/e2c8cefcf9665c1747a30e1d.png"},{"id":87849725,"identity":"83f692e7-d068-4ee5-8403-9f07d31f2066","added_by":"auto","created_at":"2025-07-29 15:46:25","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":353217,"visible":true,"origin":"","legend":"\u003cp\u003eMultivariate correlation analysis and significance test for each variable. (cardiopulmonary: cardiopulmonary diseases. Surgery time: fracture onset to surgery time. Classification: classification of hip fractures).\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-6339181/v1/98dba28743a4246830f75628.png"},{"id":87851196,"identity":"0196068b-13d2-42fc-8556-5b5b7284a528","added_by":"auto","created_at":"2025-07-29 15:54:25","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":204631,"visible":true,"origin":"","legend":"\u003cp\u003eFeature selection, ranking and error rates for random forests. A. The box-and-line plot shows the Random Forest algorithm features selected as important, tentative and unimportant variables in green, yellow and red, respectively. B. The results of the decision to reject or accept features by the Random Forest algorithm using the Boruta function run 100 times.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-6339181/v1/2d2a8789cd00a99b6aed8550.png"},{"id":87849720,"identity":"853b05ff-459e-482b-a3d6-5239daefebfb","added_by":"auto","created_at":"2025-07-29 15:46:25","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":119212,"visible":true,"origin":"","legend":"\u003cp\u003e(A). Lasso coefficient profiles created from log (lambda) series. (B). Ten-fold cross-validation of predictor variables with minimum criterion (left) and minimum plus 1 standard error (right) to determine optimal tuning parameters (lambda).\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-6339181/v1/7f0f812bc56d3231044504f0.png"},{"id":87849734,"identity":"c51864f6-de4d-4234-bca6-b7a4f0ec3347","added_by":"auto","created_at":"2025-07-29 15:46:25","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":523555,"visible":true,"origin":"","legend":"\u003cp\u003eROC curves for continuous variables in the training group, and the predictive models constructed by the Random Forest algorithm and the traditional logistic regression models in the training group are compared with each other.\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-6339181/v1/fae70b28d1f2cab0dd219f88.png"},{"id":87849738,"identity":"4cef6e0e-55d8-496a-9527-c3f51ce2f8c9","added_by":"auto","created_at":"2025-07-29 15:46:25","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":1061293,"visible":true,"origin":"","legend":"\u003cp\u003eROC curves for continuous variables in the test group, and the predictive models constructed by the Random Forest algorithm and the traditional logistic regression models in the test group are compared with each other.\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-6339181/v1/c34a94bfaea9aa42c4c5c0e2.png"},{"id":87849745,"identity":"39f5d4d1-d7fd-4e92-a97b-4506f4e112d5","added_by":"auto","created_at":"2025-07-29 15:46:26","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":109156,"visible":true,"origin":"","legend":"\u003cp\u003eMeanDecrease Accuracy and MeanDecrease Gini (important variables).\u003c/p\u003e","description":"","filename":"10.png","url":"https://assets-eu.researchsquare.com/files/rs-6339181/v1/47bed019841f3fdb4f0ab3cb.png"},{"id":98797898,"identity":"eecc7006-8825-484c-9706-8ed0df9b61f5","added_by":"auto","created_at":"2025-12-22 14:01:18","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4303634,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6339181/v1/c7ad8934-d246-422c-a6e4-17dd0e066b5c.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"The Role of Preoperative Laboratory Test Indicators in Predicting Thrombosis Risk in Elderly Hip Fracture Patients: A Random Forest Approach","fulltext":[{"header":"Introduction","content":"\u003cp\u003eWith the global aging population, osteoporosis and related conditions (such as increased fall risk) have led to a rise in the incidence of hip fractures (HF) in elderly individuals[\u003cspan additionalcitationids=\"CR2 CR3\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Hip fractures are not only a common and severe health issue among the elderly, but they also significantly increase the burden on healthcare systems[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. In addition to the direct physical consequences, elderly HF patients face a higher risk of complications, especially deep vein thrombosis (DVT) [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. After hip fractures, prolonged immobility, surgical intervention, and other factors increase the likelihood of DVT formation, which can escalate into life-threatening conditions such as pulmonary embolism (PE) [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. DVT is a common complication in HF patients, and the postoperative mortality rate in these patients is often exacerbated by DVT, making it a key issue in managing elderly patients with hip fractures[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eClinicians are increasingly concerned about preoperative DVT diagnosis and prevention, necessitating effective strategies for preoperative DVT risk assessment. Early identification of elderly hip fracture patients at high risk for preoperative or admission DVT formation is crucial. Timely and accurate diagnosis and treatment are essential to mitigate the risks of massive intraoperative pulmonary embolism, postoperative mortality, and morbidity in hip fracture patients[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eHowever, blood tests for known proinflammatory and prothrombotic states have limitations, and relying on a single indicator to accurately predict DVT formation upon admission is challenging[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Limited studies have shown that risk factors associated with preoperative DVT include postmenopausal women, previous venous thromboembolism, chronic kidney disease, delayed surgery, chronic obstructive pulmonary disease (COPD), prolonged immobilization, low mean platelet volume, abnormal serum pH, and elevated D-dimer levels[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan additionalcitationids=\"CR14\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. These risk factors are identified through single-factor analyses, highlighting the need to construct predictive models for DVT by identifying additional risk factors. As DVT diagnosis requires skilled ultrasound technicians, the training period for these specialized examiners is extended.\u003c/p\u003e\u003cp\u003eWe attempt to construct a predictive model for forecasting preoperative DVT formation by utilizing patient medical histories and characteristics available upon admission, along with necessary preoperative test indicators. The random forest predictive model can iteratively learn from new patient data, enhancing its adaptability to evolving medical information. Nowadays, machine learning, particularly the random forest algorithm\u0026mdash;a subset of machine learning algorithms\u0026mdash;has emerged as a promising tool for predicting preoperative DVT in elderly hip fracture patients[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. The random forest algorithm, known for constructing mathematical models based on sample data, has demonstrated efficacy in predicting outcomes in end-stage renal disease and type A aortic dissection (TAAD) [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. In this study, the predictive model using the random forest algorithm aims to uncover key risk factors and enhance the accuracy of preoperative DVT diagnosis in elderly hip fracture patients. By utilizing diverse data points, including the patient's physical condition, comorbidities, and clinical parameters, this research strives to construct a machine learning model tailored to the specific dynamics of elderly hip fracture cases. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e illustrates the step-by-step process of building the preoperative DVT prediction model in this study. Ultimately, the predictive modeling facilitated by the random forest algorithm lays the groundwork for improved preoperative care and a better prognosis for this aging hip fracture patient population.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eGeneral Information and Data Collection\u003c/h2\u003e\u003cp\u003eData for the two groups in this study (DVT group and non-thrombosis group) were collected from patients undergoing surgical intervention for hip fractures (HF) at West China Hospital of Sichuan University between May 2021 and November 2023. The study received approval from the Ethics Committee of West China Hospital of Sichuan University.\u003c/p\u003e\u003cp\u003eThe inclusion criteria for this study were as follows: (1) Age\u0026thinsp;\u0026ge;\u0026thinsp;60 years. (2) Patients who underwent surgical intervention during hospitalization. (3) Patients diagnosed with HF through physical examination combined with medical imaging (x-ray or computed tomography). (4) Complete clinical data.\u003c/p\u003e\u003cp\u003eThe study excluded patients with the following characteristics: (1) Secondary fractures. (2) Pre-existing hip fractures. (3) Pathologic fractures. (4) Multiple fractures or polytrauma. (5) Open fractures. (6) Old fractures (\u0026gt;\u0026thinsp;7 days). (7) Patients with poor compliance, such as those with mental illness. (8) Those undergoing conservative treatment. (9) Patients in the acute stage of cardiovascular disease. (10) Individuals without preoperative lower extremity vascular ultrasonography. (11) Patients with a history of DVT before the fracture. (12) Those involved in high-energy trauma mechanisms. (13) Cases with incomplete clinical information.\u003c/p\u003e\u003cp\u003eIn total, case data were collected from 448 elderly hip fracture patients. All collected data were analyzed anonymously, with no personal identifiers or personal privacy involved, and stored in an electronic database in Excel format. Subsequently, all cases were randomly divided into training and test datasets in a 70:30 ratio, assigning 314 patients to the training group and 134 patients to the test group.\u003c/p\u003e\u003cp\u003eClinical information for all patients was acquired from four distinct sources, encompassing demographics, chronic comorbidities, fracture-related data, and blood laboratory tests. Demographic data included patient-specific details such as gender, age, height, weight, and the calculation of Body Mass Index (BMI). Chronic comorbidities included factors related to long-standing health conditions, including alcohol consumption, smoking habits, hypertension, diabetes, and cardiopulmonary diseases. In this research, variables related to cardiopulmonary diseases primarily included Chronic Obstructive Pulmonary Disease (COPD), pulmonary infections, cardiac arrhythmias, clinically symptomatic heart failure, and coronary heart disease. Fracture-related data involved information regarding the classification of hip fractures and the duration between the occurrence of the fracture and the subsequent surgical intervention. Laboratory blood test indicators comprised the results of necessary preoperative tests, providing insights into the patient's physiological status. In this study, preoperative essential blood laboratory tests included blood cell analysis, a comprehensive coagulation panel, and biochemical blood analysis. We collected data on 63 blood indicators and, through single-factor analysis, random combinations, and correlation analysis among them, ultimately selected 22 blood indicators for inclusion in the predictive model as continuous variables. This study aimed to predict the occurrence of deep vein thrombosis (DVT) in hip fractures by gathering patient history, demographic characteristics, and readily available blood test results after admission. Laboratory biomarkers covered a comprehensive range, such as prothrombin time (PT), International Normalized Ratio (INR), fibrinogen (Fib), fibrin and fibrinogen degradation products (FDP), D-dimer, platelet count (PLT), white blood cell count (WBC), total bilirubin (TB), alanine aminotransferase (ALT), aspartate aminotransferase (AST), albumin (ALB), globulin (Glb), albumin-to-globulin ratio (A/G), creatinine (CREA), estimated glomerular filtration rate (GFR), triglyceride (TG), cholesterol (TC), creatine kinase (CK), lactate dehydrogenase (LDH), anion gap (AG), calcium (Ca), magnesium (Mg), and inorganic phosphorus (Pi). If patients underwent multiple hematological examinations after admission, we selected the first examination for analysis. The variables were also scrutinized for correlation.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eDiagnosis and management of DVT\u003c/h3\u003e\n\u003cp\u003eDeep vein thrombosis (DVT) of the lower extremities is a serious vascular condition characterized by the formation of blood clots in the deep veins of the legs. If left untreated, it can lead to life-threatening complications, such as pulmonary embolism. Early detection and management are essential.\u003c/p\u003e\u003cp\u003eThe diagnosis of DVT typically involves a comprehensive approach, including a thorough physical examination and a detailed assessment of the patient's symptoms. Ultrasound of the proximal leg veins plays a critical role in confirming the presence of DVT. This imaging technique is widely recognized and endorsed in medical guidelines, such as those provided by the American Society of Hematology[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eTreatment for DVT primarily consists of anticoagulant medications, which help prevent clot growth and reduce the risk of complications. Effective management underscores the importance of accurate diagnosis and early intervention, both of which significantly improve patient outcomes by reducing the likelihood of severe complications[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e].\u003c/p\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003eStatistical analysis\u003c/h2\u003e\u003cp\u003eFor continuous variables not conforming to a normal distribution, statistical analysis utilized nonparametric tests, and descriptive statistics employed medians (quartiles). Normal distribution conformity for continuous variables was assessed using the Kolmogorov-Smirnov method. For instance, non-paired t-tests or Wilcoxon rank-sum tests were used to compare differences between the DVT and non-DVT groups for variables like BMI and ALB in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, as well as A/G and Ca variables in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. Furthermore, based on the specific circumstances, the application of Pearson's chi-square test or Fisher's exact test was considered. Descriptive statistics, using means with positive and negative standard deviations, facilitated comparisons between groups using the independent samples t-test. Categorical variables were presented as numbers and percentages (%), and group comparisons employed the chi-square test or Fisher's exact test, as applicable.\u003c/p\u003e\u003cp\u003eThe Random Forest algorithm is a powerful machine learning method known for its accuracy and robustness in handling classification and regression tasks. Proposed by Leo Breiman in 2001, this algorithm operates by constructing a large number of decision trees. It combines the votes from each decision tree to perform effective classification or regression analyses. Each decision tree is trained on a different subset of data, and their outputs are aggregated to enhance predictive performance. The algorithm effectively captures complex relationships within the data, reducing overfitting concerns and improving generalization capabilities. Notably, it excels at reducing the risk of overfitting, effectively managing large datasets with numerous features, and highlighting the importance of individual features. The specified configuration involved the use of 500 decision trees.\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eData Preparation\u003c/strong\u003e\u003cp\u003eGender, alcohol consumption, smoking, hypertension, diabetes, classification of hip fractures, and cardiopulmonary disease were entered as dichotomous variables. Age, BMI, fracture onset to surgery time, and laboratory tests were included as continuous variables in the Random Forest program.\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eBootstrap Sampling\u003c/strong\u003e\u003cp\u003eBootstrapping and resampling/Bagging methods were used to randomly select a subset of samples for training multiple decision trees.\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eFeature Randomization\u003c/strong\u003e\u003cp\u003eA subset of m features was randomly chosen for each tree split (the Gini coefficient calculates the optimal segmentation method for m features, where m is less than the total number of features). Input variables in the risk prediction model were ranked based on the average rate of decrease in accuracy and the average rate of decrease in the Gini coefficient.\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eDecision Tree Construction\u003c/strong\u003e\u003cp\u003eIndividual decision trees were constructed using sampled data and features.\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eVoting Mechanism\u003c/strong\u003e\u003cp\u003ePredictions from all trees were aggregated to generate the final output for classification or regression.\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eEvaluating the Model\u003c/strong\u003e\u003cp\u003eThe model's performance was assessed using metrics such as precision or mean square error. In the Bagging approach, each model's training set omitted (excluded) a number of data points, constituting the Out-of-Bag (OOB) samples for each model. The OOB data was not used for training models but served as a test set. The overall estimation and generalization ability of the model's predictive performance were robustly assessed by averaging the OOB scores of all models and the OOB error. Additionally, Lasso regression was employed, considering the prominent feature selection ability of the Least Absolute Shrinkage and Selection Operator (Lasso) and the regression analysis ability of regularization. The prediction model, constructed through logistic regression, was validated on the training and test datasets by comparing the results to those of the Random Forest algorithm. Logistic regression, a linear regression technique based on the log probability ratio of the response variable, was used. Finally, the Area Under the Curve (AUC) for continuous variables and dichotomous variables was calculated. The AUC values from the random forest algorithm and traditional logistic regression were compared. The Kappa statistic and F-measure were used to test the reliability of the model. Other assessment tools included 95% confidence interval (CI), sensitivity, specificity, precision, accuracy, and balanced accuracy. The flow of the prediction model for lower extremity deep vein thrombosis in this study is depicted in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\u003c/p\u003e\u003cp\u003eAll statistical analyses in this study were conducted using RStudio (Version 2023.09.1 Build 494\u0026copy; 2009\u0026ndash;2023 Posit Software, PBC, \"Desert Sunflower\" Release [cd7011dc, 2023-10-16]) for Windows. The R software version used was 4.3.2. RStudio, along with its library packages Boruta (CRAN.R-project.org/package\u0026thinsp;=\u0026thinsp;Boruta), randomForest (CRAN.R-project.org/package\u0026thinsp;=\u0026thinsp;randomForest), and caret (CRAN.R-project.org/package\u0026thinsp;=\u0026thinsp;caret), were used for constructing random forest models. Additionally, Lasso regression was performed using RStudio and the glmnet library package. Statistical significance was determined by a p-value of \u0026lt;\u0026thinsp;0.05.\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003ePatient Characteristics\u003c/h2\u003e\u003cp\u003eIn this study, 24.55% (110/448) of the enrolled patients were elderly individuals with hip fractures (HF) and preoperative lower extremity deep vein thrombosis (DVT). Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e illustrates the step-by-step process of building the preoperative DVT prediction model in this study. The median age of the participants was 81 (71, 87), and there was a female predominance (63%). Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e displays the demographics, medical history, and laboratory test results of all enrolled patients.\u003c/p\u003e\u003cp\u003eIn the baseline comparison, statistically significant differences (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) in clinical variables were evident between the DVT and non-DVT groups. These differences included age (P\u0026thinsp;=\u0026thinsp;0.047), Fib (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), FDP (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), D-dimer (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), PLT (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), cardiopulmonary disease (P\u0026thinsp;=\u0026thinsp;0.009), and fracture onset to surgery time (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Moreover, statistically significant variations were observed in WBC (P\u0026thinsp;=\u0026thinsp;0.043), A/G (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), ALB (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), GFR (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), LDH (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and Pi (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The AUC for each variable in the univariate analysis is depicted in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFurther, D-dimer levels were 6.24 (3.91, 11.35) \u0026micro;g/mL in the non-DVT group and 9.38 (6.85, 15.01) \u0026micro;g/mL in the DVT group. PLT was 161 (123, 201) \u0026times; 10^3/\u0026micro;L in the non-DVT group and 213 (165, 266) \u0026times; 10^3/\u0026micro;L in the DVT group. ALB levels were 37.2 (31.9, 42.4) g/dL in patients without DVT and 31.75 (27.45, 35.98) g/dL in those with DVT. Fib levels were 3.33 (2.22, 4.26) g/L in the non-DVT group and 3.76 (3.16, 4.97) g/L in the DVT group.\u003c/p\u003e\u003cp\u003eAdditionally, fracture onset to surgery time was 4.2 (3.3, 5.1) days in the non-DVT group and 4.9 (4.2, 5.8) days in the DVT group. FDP was 10.1 (4.9, 22.48) \u0026micro;g/mL in the non-DVT group and 13.75 (8.45, 25.65) \u0026micro;g/mL in the DVT group. Pi was 1.04 (0.89, 1.26) mg/dL in the non-DVT group and 1.25 (1.03, 1.53) mg/dL in the DVT group.\u003c/p\u003e\u003cp\u003eSignificant variations between the non-DVT and DVT groups also included A/G, with respective values of 1.38 (1.21, 1.6) and 1.2 (1.06, 1.4); GFR, with values of 83.69 (64.97, 96.11) in the non-DVT group and 70.38 (51.98, 82.62) in the DVT group; and LDH, with values of 208 (181.25, 240) in the non-DVT group and 225 (192.25, 275.5) in the DVT group. The differences in the distribution of these laboratory indices between the DVT and non-DVT groups were statistically significant (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Figure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e displays the data visualization of the distribution of continuous variables in the DVT and non-DVT groups.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eData visualization of dichotomous variables in the non-DVT and DVT groups is presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e. Comorbidities among all enrolled patients included hypertension in 203 individuals, diabetes in 123 individuals, a history of cardiopulmonary disease in 144 individuals, a history of smoking in 173 individuals, and a history of drinking in 120 individuals. The results of the correlation analysis and significance tests for each variable are illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e. Subsequently, all cases were randomly allocated to a training dataset (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e) and a test dataset (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) in a 70:30 ratio for further analysis of their baseline characteristics.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eDemographic and clinical characteristics of study participants (Total).\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eVariables\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eTotal (n\u0026thinsp;=\u0026thinsp;448)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNo-DVT group\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eDVT group\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003ep\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003e(n\u0026thinsp;=\u0026thinsp;338,75%)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003e(n\u0026thinsp;=\u0026thinsp;110,25%)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003eDemographics\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSex, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.998\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFemale\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e283 (63)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e213 (63)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e70 (64)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMale\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e165 (37)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e125 (37)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e40 (36)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge, Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e81 (71, 87)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e81 (71, 86)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e83 (74, 88)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.047\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBMI, Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e21.92 (19.78, 24.29)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e22 (19.98, 24.34)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e21.42 (18.86, 24.18)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.161\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eMedical history\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHypertension, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.715\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e245 (55)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e187 (55)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e58 (53)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e203 (45)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e151 (45)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e52 (47)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDiabetes, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.248\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e325 (73)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e240 (71)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e85 (77)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e123 (27)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e98 (29)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e25 (23)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSmoking, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.826\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e275 (61)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e206 (61)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e69 (63)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e173 (39)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e132 (39)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e41 (37)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDrinking, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.811\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e328 (73)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e246 (73)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e82 (75)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e120 (27)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e92 (27)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e28 (25)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCardiopulmonary diseases, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.009\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e304 (68)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e241 (71)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e63 (57)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e144 (32)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e97 (29)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e47 (43)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eHip fractures\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClassification, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.196\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eExtra-articular fracture\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e247 (55)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e180 (53)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e67 (61)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIntra-articular fracture\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e201 (45)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e158 (47)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e43 (39)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFracture onset to surgery time (days)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e4.4 (3.4, 5.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e4.2 (3.3, 5.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e4.9 (4.2, 5.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMedian (Q1,Q3)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eLaboratory tests\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePT (s), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e11.7 (11.2, 12.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e11.7 (11.2, 12.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e11.8 (11.3, 12.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.751\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAPTT (s), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e28.3 (26.17, 30.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e28.3 (26.13, 30.58)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e28.1 (26.3, 30.23)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.42\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eINR, Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.07 (1.01, 1.12)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.07 (1.01, 1.12)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.07 (1.01, 1.12)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.993\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFib (g/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e3.47 (2.5, 4.47)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3.33 (2.22, 4.26)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3.76 (3.16, 4.97)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFDP (mg/mL), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e11.3 (5.7, 22.95)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e10.1 (4.9, 22.48)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e13.75 (8.45, 25.65)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eD-dimer (mg/l FEU), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e7.12 (4.23, 12.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e6.24 (3.91, 11.35)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e9.38 (6.85, 15.01)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRBC (\u0026times;10^12/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e3.59 (3.01, 4.16)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3.64 (3.08, 4.17)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3.42 (2.94, 4.04)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.12\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePLT (\u0026times;10^9/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e171 (133.75, 223.75)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e161 (123, 201)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e213 (165, 266)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWBC (\u0026times;10^9/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e8.54 (6.69, 11.14)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e8.46 (6.58, 10.56)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e8.82 (7, 12.46)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.043\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALT (U/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e14 (10, 21)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e14 (11, 21)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e14 (10, 22)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.954\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAST (U/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e20.5 (17, 26)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e20 (17, 25)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e21 (17, 31)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.309\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALB (g/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e35.9 (30.48, 40.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e37.2 (31.9, 42.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e31.75 (27.45, 35.98)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGlb (g/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e24.7 (21.8, 27.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e24.7 (21.92, 27.67)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e24.1 (21.8, 28.12)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.544\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eA/G, Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.35 (1.18, 1.56)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.38 (1.21, 1.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.2 (1.06, 1.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCREA (umol/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e92 (76, 123)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e90.5 (76, 117.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e93 (73, 165.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.409\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGFR (mL/min/1.73m\u0026sup2;), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e79.47 (61.44, 93.64)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e83.69 (64.97, 96.11)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e70.38 (51.98, 82.62)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTG (mmol/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.05 (0.82, 1.38)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.06 (0.84, 1.37)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.01 (0.71, 1.43)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.258\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTC (mmol/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e3.46 (2.88, 3.92)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3.46 (2.87, 3.95)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3.47 (2.88, 3.92)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.852\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCK (U/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e95 (55.75, 207.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e91 (54.25, 179.75)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e114 (58, 264.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.057\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLDH (U/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e212 (185, 249.25)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e208 (181.25, 240)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e225 (192.25, 275.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCa(mmol/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2.14 (2.06, 2.23)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2.14 (2.06, 2.23)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2.14 (2.06, 2.23)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.795\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePi(mmol/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.08 (0.91, 1.35)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.04 (0.89, 1.26)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.25 (1.03, 1.53)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eMedian (Q1, Q3): Continuous variables that did not fit the normal distribution used the median (quartiles). Mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD: Continuous variables that fit the normal distribution were tested to obtain the mean (plus or minus standard deviation, SD) using the Kolmogorov-Smirnov method. n (%): Categorical variables expressed as numbers and percentages (%).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eDemographic and clinical characteristics of training group participants.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eVariables\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eTrains (n\u0026thinsp;=\u0026thinsp;314)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNo-DVT group\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eDVT group\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003ep\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003e(n\u0026thinsp;=\u0026thinsp;237,75%)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003e(n\u0026thinsp;=\u0026thinsp;77,25%)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003eDemographics\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSex, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.741\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFemale\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e201 (64)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e150 (63)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e51 (66)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMale\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e113 (36)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e87 (37)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e26 (34)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge, Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e81 (72, 88)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e81 (71, 87)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e83 (74, 89)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.083\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBMI, Mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e22.16\u0026thinsp;\u0026plusmn;\u0026thinsp;3.5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e22.22\u0026thinsp;\u0026plusmn;\u0026thinsp;3.35\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e21.98\u0026thinsp;\u0026plusmn;\u0026thinsp;3.96\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.638\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eMedical history\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHypertension, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e176 (56)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e133 (56)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e43 (56)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e138 (44)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e104 (44)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e34 (44)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDiabetes, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.76\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e222 (71)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e166 (70)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e56 (73)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e92 (29)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e71 (30)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e21 (27)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSmoking, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.393\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e193 (61)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e142 (60)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e51 (66)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e121 (39)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e95 (40)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e26 (34)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDrinking, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.736\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e234 (75)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e175 (74)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e59 (77)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e80 (25)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e62 (26)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e18 (23)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCardiopulmonary diseases, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.017\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e216 (69)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e172 (73)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e44 (57)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e98 (31)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e65 (27)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e33 (43)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eHip fractures\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClassification, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.316\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eExtra-articular fracture\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e170 (54)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e124 (52)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e46 (60)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIntra-articular fracture\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e144 (46)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e113 (48)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e31 (40)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFracture onset to surgery time (days)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e4.4 (3.4, 5.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e4.2 (3.2, 5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e5 (4.3, 5.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMedian (Q1,Q3)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eLaboratory tests\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePT (s), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e11.8 (11.3, 12.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e11.8 (11.3, 12.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e11.8 (11.2, 12.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.665\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAPTT (s), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e28.15 (26, 30.28)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e28.2 (26, 30.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e28.1 (26, 30.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.919\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eINR, Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.07 (1, 1.12)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.06 (1, 1.12)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.07 (1, 1.12)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.733\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFib (g/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e3.43 (2.48, 4.41)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3.32 (2.18, 4.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3.75 (3.14, 4.87)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFDP (mg/mL), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e11 (5.8, 22.58)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e9 (5, 21.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e15.3 (8.7, 24.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eD-dimer (mg/l FEU), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e7.32 (4.23, 12.14)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e6.26 (3.85, 11.36)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e9.77 (7.29, 13.54)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRBC (\u0026times;10^12/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e3.58 (3.03, 4.15)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3.65 (3.13, 4.17)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3.34 (2.89, 3.97)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.029\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePLT (\u0026times;10^9/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e169 (135, 231)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e161 (123, 198)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e215 (160, 259)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWBC (\u0026times;10^9/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e8.68 (6.71, 11.33)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e8.52 (6.59, 10.99)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e8.84 (7.46, 12.41)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.124\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALT (U/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e14 (11, 21)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e15 (11, 21)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e14 (10, 24)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.838\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAST (U/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e21 (17, 27)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e21 (17, 25)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e21 (17, 37)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.164\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALB (g/L), Mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e35.47\u0026thinsp;\u0026plusmn;\u0026thinsp;7.72\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e36.9\u0026thinsp;\u0026plusmn;\u0026thinsp;7.69\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e31.05\u0026thinsp;\u0026plusmn;\u0026thinsp;6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGlb (g/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e24.7 (21.8, 27.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e24.7 (21.8, 27.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e24.1 (21.8, 29.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.684\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eA/G, Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.34 (1.17, 1.56)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.37 (1.21, 1.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.19 (1.07, 1.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCREA (umol/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e92.5 (76, 126)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e89 (76, 116)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e96 (73, 167)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.292\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGFR (mL/min/1.73m\u0026sup2;), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e78.75 (60.43, 93.08)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e83.19 (63.02, 95.47)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e69.19 (51.89, 80)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTG (mmol/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.09 (0.8, 1.43)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.13 (0.84, 1.43)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.97 (0.69, 1.38)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.054\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTC (mmol/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e3.46 (2.88, 3.88)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3.48 (2.88, 3.89)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3.45 (2.82, 3.78)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.718\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCK (U/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e101 (58, 234.25)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e101 (57, 197)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e121 (59, 287)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.109\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLDH (U/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e214.5 (186.25, 251)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e208 (182, 244)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e225 (196, 270)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.006\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCa(mmol/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2.14 (2.07, 2.23)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2.14 (2.07, 2.23)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2.14 (2.06, 2.23)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.772\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePi(mmol/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.08 (0.89, 1.34)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.05 (0.89, 1.27)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.23 (1.03, 1.46)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eMedian (Q1, Q3): Continuous variables that did not fit the normal distribution used the median (quartiles). Mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD: Continuous variables that fit the normal distribution were tested to obtain the mean (plus or minus standard deviation, SD) using the Kolmogorov-Smirnov method. n (%): Categorical variables expressed as numbers and percentages (%).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eDemographic and clinical characteristics of test group participants.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eVariables\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eTest (n\u0026thinsp;=\u0026thinsp;134)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNo-DVT group\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eDVT group\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003ep\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003e(n\u0026thinsp;=\u0026thinsp;101,75%)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003e(n\u0026thinsp;=\u0026thinsp;33,25%)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003eDemographics\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSex, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.775\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFemale\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e82 (61)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e63 (62)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e19 (58)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMale\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e52 (39)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e38 (38)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e14 (42)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge, Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e80 (70.25, 85)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e79 (70, 84)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e82 (71, 86)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.28\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBMI, Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e21.48 (19.05, 23.78)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e21.83 (20.03, 24)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e20.54 (18.36, 22.94)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.175\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eMedical history\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHypertension, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.549\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e69 (51)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e54 (53)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e15 (45)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e65 (49)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e47 (47)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e18 (55)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDiabetes, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.136\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e103 (77)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e74 (73)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e29 (88)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e31 (23)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e27 (27)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e4 (12)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSmoking, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.486\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e82 (61)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e64 (63)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e18 (55)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e52 (39)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e37 (37)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e15 (45)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDrinking, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e94 (70)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e71 (70)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e23 (70)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e40 (30)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e30 (30)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e10 (30)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCardiopulmonary diseases, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.359\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e88 (66)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e69 (68)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e19 (58)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eYse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e46 (34)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e32 (32)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e14 (42)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eHip fractures\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClassification, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.533\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eExtra-articular fracture\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e77 (57)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e56 (55)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e21 (64)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIntra-articular fracture\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e57 (43)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e45 (45)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e12 (36)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFracture onset to surgery time (days)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e4.3 (3.43, 5.27)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e4.2 (3.4, 5.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e4.7 (4.2, 5.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMedian (Q1,Q3)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eLaboratory tests\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePT (s), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e11.5 (11.1, 12.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e11.4 (11.1, 12.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e11.6 (11.3, 12.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.153\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAPTT (s), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e28.7 (26.72, 31.15)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e28.9 (26.9, 31.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e27.9 (26.7, 29.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.081\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eINR, Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.07 (1.03, 1.13)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.07 (1.03, 1.13)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.06 (1.01, 1.13)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.576\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFib (g/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e3.49 (2.55, 4.61)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3.34 (2.47, 4.35)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3.81 (3.17, 5.17)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.02\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFDP (mg/mL), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e11.9 (5.62, 24.53)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e11.3 (4.7, 23.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e12.2 (8.2, 25.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.184\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eD-dimer (mg/l FEU), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e6.76 (4.31, 12.58)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e6.16 (4.03, 10.36)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e8.35 (6.69, 19.42)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRBC (\u0026times;10^12/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e3.62 (3, 4.22)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3.61 (3.05, 4.21)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3.64 (3, 4.28)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.668\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePLT (\u0026times;10^9/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e172 (133.25, 212.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e160 (122, 201)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e200 (172, 278)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWBC (\u0026times;10^9/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e8.42 (6.6, 10.71)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e8.32 (6.48, 10.23)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e8.73 (6.78, 12.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.184\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALT (U/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e13 (10, 19)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e13 (10, 19)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e14 (10, 18)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.8\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAST (U/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e20 (16, 25)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e20 (16, 25)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e20 (17, 23)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.784\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALB (g/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e37.05 (30.55, 41.38)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e38.3 (32.3, 43.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e32.7 (27.7, 36.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGlb (g/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e24.65 (22, 27.17)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e24.7 (22, 27.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e24.2 (21.6, 26.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.657\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eA/G, Mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.37\u0026thinsp;\u0026plusmn;\u0026thinsp;0.33\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.41\u0026thinsp;\u0026plusmn;\u0026thinsp;0.32\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.22\u0026thinsp;\u0026plusmn;\u0026thinsp;0.31\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.004\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCREA (umol/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e87 (69, 123)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e91 (72, 119)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e79 (69, 153)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.92\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGFR (mL/min/1.73m\u0026sup2;), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e82.46 (65.51, 95.64)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e87.09 (66.79, 97.86)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e74 (52.23, 88.89)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.022\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTG (mmol/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.04 (0.85, 1.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.02 (0.85, 1.28)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.21 (0.85, 1.52)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.242\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTC (mmol/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e3.46 (2.87, 4.28)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3.45 (2.8, 4.18)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3.52 (3.11, 4.31)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.303\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCK (U/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e85.5 (48.25, 143)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e83 (48, 135)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e101 (51, 206)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.271\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLDH (U/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e209.5 (181.25, 246.75)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e209 (181, 236)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e217 (185, 289)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.057\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCa(mmol/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2.13\u0026thinsp;\u0026plusmn;\u0026thinsp;0.13\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2.14\u0026thinsp;\u0026plusmn;\u0026thinsp;0.13\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2.13\u0026thinsp;\u0026plusmn;\u0026thinsp;0.15\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.694\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePi(mmol/L), Mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.06 (0.95, 1.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.03 (0.93, 1.25)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.36 (1.06, 1.56)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eMedian (Q1, Q3): Continuous variables that did not fit the normal distribution used the median (quartiles). Mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD: Continuous variables that fit the normal distribution were tested to obtain the mean (plus or minus standard deviation, SD) using the Kolmogorov-Smirnov method. n (%):Categorical variables expressed as numbers and percentages (%).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eFeature Screening\u003c/h2\u003e\u003cp\u003eIn this study, the random forest prediction model was employed to meticulously examine each variable for its potential to predict the occurrence of preoperative deep vein thrombosis (DVT) in elderly patients with hip fractures (HF). The process began with random forest-based feature screening, as depicted in Figs.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA and B, which illustrated both the process and outcomes of feature selection via the random forest algorithm. From these results, 11 key attributes were identified as crucial for predicting the target variable (DVT). These attributes are ALB (albumin), D-dimer, PLT (platelet count), Fib (fibrinogen), A/G (albumin-to-globulin ratio), Pi (inorganic phosphorus), CREA (creatinine), GFR (estimated glomerular filtration rate), FDP (fibrin and fibrinogen degradation products), cardiopulmonary disease, and fracture onset to surgery time. Additionally, three tentative attributes\u0026mdash;AST (aspartate aminotransferase), sex, LDH (lactate dehydrogenase), and TG (triglyceride)\u0026mdash;were identified and require further investigation due to uncertain importance. By utilizing Boruta analysis for feature selection, the model was streamlined, focusing on the most relevant variables to enhance prediction accuracy and efficiency.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eWe then identified the feature selection results of the Random Forest algorithm through the Lasso regression model. The Lasso coefficient distributions of the features and the optimal penalty coefficients (λ) are shown in Fig.\u0026nbsp;7A-B. Unsurprisingly, due to the strong contraction and regularized regression analysis ability of Lasso regression, only 9 variables were left in the Lasso regression model associated with the occurrence of preoperative DVT in elderly HF patients. This set of variables completely overlapped with the important attribute variables identified in Random Forest.\u003c/p\u003e\u003cp\u003eSubsequently, we leveraged the Lasso regression model to validate the feature selection outcomes derived from the Random Forest algorithm. As depicted in Fig.\u0026nbsp;7A-B, the distributions of Lasso coefficients for the features and the optimal penalty coefficients (λ) are presented. Given the inherent capacity of Lasso regression to impose strong constraints and conduct regularized regression analysis, it comes as no surprise that only nine variables remained in the Lasso regression model, which are associated with the occurrence of preoperative DVT in elderly patients with hip fractures (HF). Remarkably, this set of variables fully coincided with the important attribute variables identified through Random Forest, thereby reinforcing the robustness and reliability of the selected features.Figure 7. (A). Lasso coefficient profiles created from log (lambda) series. (B). Ten-fold cross-validation of predictor variables with minimum criterion (left) and minimum plus 1 standard error (right) to determine optimal tuning parameters (lambda).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eRandom Forest Prediction Model Construction and Validation\u003c/h3\u003e\n\u003cp\u003eFollowing the identification of the 11 crucial risk factors through the Random Forest algorithm, we proceeded to construct a risk prediction model. The model was built using these key variables, and the results indicated that a random forest classification model with a 500-class decision tree, which attempts four variables at each segmentation, performed effectively. The out-of-bag (OOB) error rate was calculated to be 13.69%, signifying a relatively small generalization error and suggesting that the model has good predictive performance.\u003c/p\u003e\u003cp\u003eThe Receiver Operating Characteristic (ROC) curves for continuous variables, predictive models constructed by the random forest algorithm, and traditional logistic regression were visualized in both the training dataset (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003e) and the test dataset (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e9\u003c/span\u003e). The area under the curve (AUC) for the prediction model based on the random forest algorithm was 1.000 in the training dataset and 0.899 in the test dataset, indicating excellent discrimination performance. In contrast, the AUC values for the risk prediction model constructed by logistic regression were 0.885 in the training dataset and 0.814 in the test dataset. The differences in AUC values between the random forest algorithm prediction model and the logistic regression risk prediction model highlight the superior discriminative performance of the random forest algorithm prediction model.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe importance of the 11 variables in the Random Forest categorical prediction model of this study is detailed in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e10\u003c/span\u003e. The results of the DVT group columns, indicating the importance of the variables, highlight fracture onset to surgery time with the highest value at 22.611865, signifying its paramount importance in this DVT prediction model. In contrast, cardiopulmonary diseases have the lowest value of 6.535452, suggesting relatively lesser importance for DVT prediction among these 11 variables.\u003c/p\u003e\u003cp\u003eThe mean decrease accuracy column reveals that excluding the fracture onset to surgery time variable results in a significant 27.93% reduction in model accuracy, emphasizing its key role in accuracy. On the other hand, the mean reduced Gini coefficient indicates that higher values of the variable are more effective at segmenting the data. In this predictive model, fracture onset to surgery time emerges as the most crucial variable, playing a pivotal role in accurately predicting the occurrence of DVT. Additionally, D-dimer, PLT, and ALB are also important factors in predicting DVT occurrence. Overall, this Random Forest model heavily relies on ALB, PLT, and fracture onset to surgery time for accurate predictions.\u003c/p\u003e\u003cp\u003eThe validation results reveal that the method demonstrates strong discriminative properties, achieving impeccable classification across various metrics. In the training set, the 95% Confidence Interval (CI), Sensitivity, Specificity, Precision, Accuracy, and Balanced Accuracy stand at (0.9883, 1), 1.000, 1.000, 1.000, 1.000, and 1.000, respectively. For the test set, the 95% CI, Sensitivity, Specificity, Precision, Accuracy, and Balanced Accuracy are (0.7875, 0.9124), 0.6061, 0.9406, 0.7692, 0.8582, and 0.7733, respectively. This comprehensive evaluation underscores the method's robust performance and its ability to maintain high accuracy and balanced metrics in both training and test scenarios.\u003c/p\u003e\u003cp\u003eThe reliability of the predictive model was assessed using the Kappa statistic and F1 Score. In the training dataset, the Kappa values for the random forest-based prediction model and the logistic regression-based prediction model were 1.000 and 0.4815, respectively. In the test dataset, these values were 0.5887 and 0.3761, respectively. Examining the F1 Score, the random forest-based prediction model achieved a score of 1.000 in the training dataset, while the logistic regression-based prediction model scored 0.6408. In the test dataset, the F1 scores were 0.6780 for the random forest model and 0.5682 for the logistic regression model. These metrics provide insights into the model's reliability, with the random forest model consistently outperforming the logistic regression model across both training and test datasets.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eImportance, MeanDecrease Accuracy and MeanDecrease Gini (important variables).\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVariables\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNo-DVT group\u003c/p\u003e\u003cp\u003eImportance\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eDVT group\u003c/p\u003e\u003cp\u003eImportance\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eMeanDecrease Accuracy\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eMeanDecrease Gini\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCardiopulmonary\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e7.761557\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e6.535452\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e9.311122\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e2.524233\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFracture onset to surgery time\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e19.131834\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e27.935109\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e27.935109\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e20.649274\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFib\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e6.805638\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e8.437195\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e9.815045\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e8.929377\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFDP\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e4.251672\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e6.606914\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e7.641734\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e6.653149\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eD-dimer\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e12.040965\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e11.216333\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e15.779695\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e12.408828\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePLT\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e6.325702\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e13.072805\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e13.072691\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e11.943518\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e7.765926\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e20.678917\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e18.478731\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e14.203333\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eA/G\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e6.176767\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e6.718918\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e8.463501\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e11.171071\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCREA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e6.496627\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e8.778142\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e10.283946\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e8.467382\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGFR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e4.126512\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e6.750375\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e7.565805\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e9.319140\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePi\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e6.481033\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e7.447505\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e9.497667\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e9.349841\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eWith the aging of the world's population, there is a growing interest in managing elderly HF patients[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. Factors such as age, gender, low serum vitamin D levels, and low bone mineral density are significantly associated with osteoporosis[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. The elderly population is more susceptible to HFs. DVT often develops in elderly patients suffering HFs due to limited mobility, vascular injury, altered coagulation factors post-fracture, surgery, and diabetic comorbidities[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. DVT is more perilous than other HF complications as it can lead to life-threatening diseases like pulmonary embolism, posing a threat to the lives of elderly patients[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. However, diagnosing DVT in elderly patients with fractures is limited by the prevalence of comorbidities that may mask symptoms and complicate accurate assessments. Predicting DVT in elderly HF patients soon after hospital admission is crucial for early intervention, personalized treatment plans, and reducing the risk of postoperative mortality[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. It also enhances clinical prognosis. While current DVT prediction models often amalgamate risk factors using methods like univariate regression followed by multivariate logistic regression, they exhibit limitations in accurately predicting outcomes for individual patients in complex elderly cases. In response, we developed a robust random forest prediction model, demonstrating significant discriminatory power for predicting deep vein thrombosis. The ROC visualization analysis underscored the superior predictive accuracy of the random forest algorithm over classical logistic regression in both the training and test datasets.\u003c/p\u003e\u003cp\u003eRandom forest prediction models are ensemble learning algorithms that construct a multitude of decision trees during the training phase and subsequently leverage these trees for making predictions[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Each decision tree is trained on a distinct subset of the data, and the model consolidates the individual predictions to arrive at a final decision. The unique strength of the Random Forest model lies in its capacity to manage large datasets with high dimensionality, even encompassing thousands of clinical variables[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. It achieves this by mitigating overfitting through the averaging of predictions from multiple trees and by providing rankings of feature importance. The advantages of random forest models over other models include robustness to noisy data, the ability to handle both classification and regression tasks, and resistance to overfitting, ultimately leading to enhanced generalization performance[\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e].Our study marks the first attempt to integrate the Random Forest algorithm into predicting admission thrombosis in elderly hip fracture (HF) patients. This integration aims to develop a prediction model using basic characteristics of admitted HF patients, their medical history, and essential preoperative laboratory blood test indicators. The Random Forest prediction model has demonstrated significant potential for foreseeing thrombosis in newly admitted elderly HF patients. As machine learning, particularly the Random Forest algorithm, continues to evolve and validate, coupled with the abundance of clinical data, the exploration and development of Random Forest predictive models for various diseases warrant considerable attention[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e].Compared to traditional logistic regression, random forests are less sensitive to multicollinearity and exhibit relative robustness to missing data[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. The predictive results of the random forest algorithm remain stable and are less influenced by existing missing data compared to logistic regression. With the advancement of big data, the prevalence of multicenter clinical studies continues to rise. In scenarios involving several thousand variables, logistic regression proves inadequate, while the random forest algorithm can offer more accurate predictions.\u003c/p\u003e\u003cp\u003eIn this study, we found that 25% of elderly hip fracture (HF) patients who underwent surgical intervention during hospitalization developed deep vein thrombosis (DVT). The prediction model focused on easily obtainable clinical variables to enable swift and efficient prediction. The Random Forest algorithm highlighted specific variables, such as ALB (albumin), A/G (albumin-to-globulin ratio), PLT (platelet count), Fib (fibrinogen), D-dimer, CREA (creatinine), Pi (inorganic phosphorus), GFR (estimated glomerular filtration rate), FDP (fibrin and fibrinogen degradation products), fracture onset to surgery time, and cardiopulmonary diseases, as valuable for characterizing and predicting preoperative thrombosis in elderly Chinese HF patients during hospitalization. Subsequently, we validated these nine variables using Lasso regression modeling. In previous studies, age at surgery has been recognized as a significant risk factor for preoperative DVT in HF patients, particularly among the elderly compared to other age groups. In Wang et al.'s study, surgical age was classified into four groups: 60\u0026ndash;70 years, 70\u0026ndash;80 years, 80\u0026ndash;90 years, and \u0026gt;\u0026thinsp;90 years[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. The results indicated a heightened risk of preoperative DVT in patients over 90 years, with no statistically significant differences among the other age groups[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. This observation may explain the lack of significance of age as a variable in the Random Forest predictive model used in this study. The study focused on five laboratory test indicator variables, which have been previously documented in the literature.\u003c/p\u003e\u003cp\u003eIn the investigation exploring the correlation between cardiovascular diseases and the formation of deep vein thrombosis (DVT), several significant studies have been conducted. Gregson et al.[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e] carried out a prospective cohort study to examine the association between cardiovascular risk factors and venous thromboembolism (VTE), which includes DVT and pulmonary embolism (PE). They identified older age, smoking, and obesity as overlapping risk factors for both conditions. Liu et al.'s multicenter cohort study revealed a 24.5% incidence of venous thromboembolism (VTE) in patients with acute exacerbation of chronic obstructive pulmonary disease (AECOPD) [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. The major risk factors identified included older age, higher body mass index, and a history of VTE. An interesting finding emerged regarding the impact of DVT formation on acute cardiopulmonary symptoms, including asymptomatic pulmonary embolism. Hou et al.[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e] investigated the association between DVT and pulmonary embolism (PE). Their results indicated that the occurrence of PE in patients with lower extremity deep vein thrombosis is associated with alcohol consumption and heart failure. The proximal veins were more affected than the distal ones, and the right side was more affected than the left. Numerous studies have demonstrated the biological mechanisms by which cardiovascular diseases lead to deep vein thrombosis (DVT)[\u003cspan additionalcitationids=\"CR31\" citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]: Hemodynamic changes: On one hand, conditions such as heart failure can result in reduced cardiac output, thereby causing blood flow to slow down. This stasis can promote the formation of deep vein thrombosis. On the other hand, prolonged inactivity or bed rest, which is common in patients with hip fractures, can lead to venous stasis. This increases the risk of DVT as blood accumulates in the veins. Endothelial dysfunction: On one hand, cardiovascular diseases such as atherosclerosis can damage the inner layer of blood vessels. This damage exposes the underlying tissue, activates the coagulation cascade, and promotes thrombosis. On the other hand, chronic inflammation associated with cardiovascular diseases can impair endothelial function. Inflammatory markers such as C-reactive protein (CRP) and interleukin-6 (IL-6) can promote a pre-thrombotic state by upregulating adhesion molecules and procoagulant factors on endothelial cells. Hypercoagulable state: Cardiovascular diseases can lead to an increase in coagulation factors, such as fibrinogen, factor VII, and factor VIII. These factors enhance the coagulation process and increase the risk of thrombosis. Additionally, obesity, which is a risk factor for cardiovascular diseases, can cause an increase in the levels of plasminogen activator inhibitor-1 (PAI-1), inhibiting the conversion of plasminogen to plasmin and thus reducing the breakdown of fibrin clots, leading to a decrease in the activity of the fibrinolytic system.\u003c/p\u003e\u003cp\u003eThe impact of the time from fracture occurrence to surgical intervention on the incidence of deep vein thrombosis (DVT) has been extensively studied. Yang et al.'s research emphasizes the critical role of delayed time from injury to surgery in increasing the risk of preoperative deep vein thrombosis in fracture patients[\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. This retrospective study underscores the importance of timely surgical intervention, particularly in elderly patients, for preventing DVT. In a study focusing on hip fractures published in the \"Journal of Orthopaedic Surgery and Research,\" Taoka et al.[\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e] found that a delay of more than 48 hours in surgical intervention after hip fracture significantly increases the risk of developing proximal DVT. Prolonged bed rest leads to slow blood flow in the lower limbs, especially in the deep veins. Blood stasis provides an opportunity for thrombus formation. At the same time, due to hip fractures, there is a lack of muscle activity in the legs, which normally helps to propel blood upward into the veins, leading to blood pooling. The longer the delay in surgery, the more pronounced these hemodynamic changes become, thereby increasing the risk of DVT.\u003c/p\u003e\u003cp\u003eWhen a blood clot forms in the body, the coagulation system is activated, leading to the conversion of fibrinogen to fibrin. The fibrin forms a mesh-like structure to stabilize the clot. Subsequently, the fibrinolytic system is also activated to break down the fibrin clot. D-dimer is a specific degradation product of cross-linked fibrin. Its presence and increased levels in the blood indicate that there has been recent fibrin formation and subsequent degradation, which is a key process in thrombosis. Elevated D-dimer levels, coupled with indications of an active fibrinolytic system, suggest the potential presence of a clot, thereby increasing the risk of thrombotic events such as DVT in elderly patients with hip fractures (HF)[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Some studies have noted age-dependent variations in the specificity of D-dimer, with reduced accuracy, particularly in individuals over 65 years old. In a retrospective study, D-dimer and fibrinogen levels emerged as potential predictors of postoperative thrombosis in patients with lower extremity fractures[\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. Beyond D-dimer, other coagulation function indicators, such as Fib and FDP, have also emerged as potential biomarkers for thrombus formation[\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. For instance, Bai et al.[\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e] investigated thrombotic risk prediction in patients with acute promyelocytic leukemia (APL) and observed significantly elevated levels of FDP and D-dimer in APL patients with thrombosis. The researchers proposed that the ratios of FDP/Fib and D-dimer/Fib could be significantly correlated factors in predicting thrombus formation. These ratios also hold value as prognostic indicators, particularly in high-risk patients prone to thrombosis. Similarly, Pang et al.[\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e] identified elevated levels of D-dimer and fibrinogen degradation products in patients with post-mastectomy DVT, emphasizing their importance as valuable indicators of thrombus formation. Additionally, Yang et al.'s research indicated that FDP is an independent risk factor and a significant predictor of deep vein thrombosis in patients with acute coronary syndrome, underscoring its importance as a valuable marker in thrombus formation[\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eAlbumin (ALB), a major plasma protein, plays a crucial role in maintaining vascular integrity and preventing thrombosis. Studies have shown[\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e] that ALB can bind with arachidonic acid, preventing its metabolism into potent aggregating substances, endoperoxides, and thromboxane A2, thereby inhibiting platelet activation and aggregation. In addition, it exhibits a concentration-dependent ability to induce inducible nitric oxide synthase in macrophages, thereby increasing the production of nitric oxide, a potent platelet inhibitor. In the presence of inflammation, increased exertion, or injury, ALB levels typically decrease, making it a potential predictor of deep vein thrombosis (DVT) in elderly patients with hip fractures (HFs). Despite ALB's dependence on various physiological factors and health status, caution is warranted in interpreting it as an independent predictor. However, the random forest prediction model employed in this study allows for analytical prediction by integrating diverse variables. Wu et al.[\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e] conducted a retrospective examination of the correlation between serum albumin levels and preoperative DVT in patients aged 65 years and older with HF. The results indicated a 6% decrease in DVT risk for every 1 g/L increase in albumin concentration after accounting for confounding factors, suggesting its potential as a predictor of DVT risk in the elderly. Yao et al.[\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e] utilized the D-dimer to albumin ratio (DAR) to predict perioperative DVT in elderly patients with HF, achieving a final AUC of 0.677.\u003c/p\u003e\u003cp\u003ePlatelets are one of the main participants in the processes of thrombosis formation and hemostasis[\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. Elevated platelet levels have the potential to initiate the coagulation cascade and enhance platelet self-aggregation, making them potential predictors of deep vein thrombosis (DVT). In the Nomogram prediction model for preoperative DVT in the elderly developed by Zhang et al.[\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e], the platelet (PLT) count emerged as one of the independent risk factors, showing a statistical difference between groups. Similarly, a PLT count of 220 \u0026times; 10^9/L was identified as an independent factor for DVT in a study by Niu et al.[\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e]. The risk of DVT in elderly hip fracture (HF) patients with preoperative levels exceeding 220 \u0026times; 10^9/L was 2.02 times higher than in patients without DVT. Wang et al.[\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e] demonstrated a significant correlation, indicating that a mean platelet volume (MPV) level\u0026thinsp;\u0026lt;\u0026thinsp;13.3 fL was associated with DVT.\u003c/p\u003e\u003cp\u003eCreatinine (CREA) serves as an indicator of renal function, and its elevation typically arises due to impaired renal function. Studies have shown[\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e] that elevated creatinine levels are associated with increased levels of coagulation factors (such as FVIII), which can lead to a hypercoagulable state in the blood and increase the risk of thrombosis. When kidney function is impaired, changes in hemodynamics may affect the pressure of blood return in the lower limbs, leading to stasis of blood in the lower limb veins. On the other hand, the levels of metabolic waste and toxins in the blood (such as urea and creatinine) will increase, which may lead to an increase in blood viscosity. A creatinine level exceeding 0.96 mg/dL has been identified as a significant predictor of thromboembolic events, as demonstrated in a multifactorial analysis within a nested study conducted by Haltout et al.[\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e]. In a separate study, Ding et al. developed an XGBoost model for machine learning screening and predicting risk factors associated with deep vein thrombosis (DVT) following hip arthroplasty. Notably, CREA was integrated as a primary characterization indicator in this predictive model. The presence of chronic kidney disease (CKD) contributes to a decline in glomerular filtration rate (GFR), potentially inducing hemostatic changes that could influence the risk of DVT formation. In the groundbreaking study by Wang et al.[\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e], the first exploration of the relationship between chronic kidney dysfunction and preoperative DVT in hip fracture patients was conducted. Among patients with an estimated GFR (eGFR) less than 60 mL/min/1.73 m\u0026sup2;, the incidence of DVT was 6.0%. The study establishes that eGFR serves as an independent predictive factor for preoperative DVT formation. Additionally, elevated fibrinogen levels were identified as another independent predictive factor for DVT formation. Li et al.'s research revealed a 2.68-fold increase in the risk of DVT in stage 3/4 CKD patients[\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e]. Severe chronic kidney disease may render patients more susceptible to proximal and symptomatic deep vein thrombosis. CKD emerges as a significant risk factor for DVT occurrence after total hip arthroplasty.\u003c/p\u003e\u003cp\u003eOrganic phosphorus is involved in glycolysis, ammonification, and oxidative phosphorylation, generating chemical energy by producing adenosine triphosphate (ATP) from adenosine diphosphate (ADP). It also affects the oxygen-carrying capacity of hemoglobin by regulating the synthesis of 2,3-diphosphoglycerate. Phosphorus atoms are components of DNA and RNA bases, as well as phospholipids, which are involved in cell structure and signal transduction[\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e]. Therefore, a decrease in blood phosphorus levels in the human body can lead to cell damage, tissue hypoxia, and multi-organ damage, especially in vascular endothelial cells, causing the aggregation of platelets and coagulation factors. Elevated blood phosphorus levels are also an important factor in the occurrence of deep vein thrombosis (DVT)[\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e]. When blood phosphorus levels increase, beta-glycerophosphate in human endothelial cells induces endothelial cell damage by increasing the expression of p16 and the activity of age-related beta-galactosidase. Lu et al.[\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e] used serum phosphorus as a continuous variable. They identified risk factors for DVT following hip arthroplasty through a comprehensive analysis involving multivariate binary logistic regression and generalized additive models. Their research findings support the view that serum phosphorus levels can serve as a predictive indicator of the risk of developing DVT.\u003c/p\u003e\u003cp\u003eOur study has several limitations. Firstly, it is retrospective in nature, which may introduce selection bias and limit the ability to establish causality. Secondly, our focus was solely on the incidence of lower extremity deep vein thrombosis (DVT), excluding other sites, which may not provide a comprehensive view of the overall thrombotic risk. Thirdly, our reliance on the quality of medical record data hindered patient validation, potentially affecting the accuracy of the results. Lastly, our study population consisted exclusively of elderly hip fracture (HF) patients undergoing surgical intervention during hospitalization, rather than encompassing all elderly HF patients, which resulted in an insufficiently large sample size. To enhance the accuracy of the random forest prediction model, future research should involve extensive multicenter samples to increase the sample size and improve the generalizability of the findings. Additionally, a prospective study design could help mitigate some of the limitations associated with retrospective data collection. Including a broader range of patients, such as those with different types of fractures or those managed conservatively, would provide a more comprehensive understanding of the risk factors for DVT. Furthermore, incorporating more detailed and accurate data collection methods, possibly through direct patient interviews or more rigorous data entry protocols, could enhance the quality of the data and the robustness of the model.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eIn this study, we successfully developed and validated a Random Forest-based predictive model to anticipate preoperative deep vein thrombosis (DVT) in elderly hip fracture (HF) patients. The model demonstrated robust discriminative performance, achieving an AUC of 1.000 in the training set and 0.899 in the test set, significantly outperforming traditional logistic regression approaches. Key preoperative indicators\u0026mdash;including ALB, D-dimer, PLT, fracture-to-surgery time, and cardiopulmonary comorbidities\u0026mdash;were identified as critical predictors, aligning with pathophysiological mechanisms of thrombosis. This model not only enhances preoperative risk stratification but also provides clinicians with actionable insights to implement timely interventions, such as early anticoagulation or optimized surgical scheduling, thereby reducing perioperative morbidity and mortality. Future research will focus on expanding the sample size and refining the model to further improve its predictive accuracy and clinical utility.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003cstrong\u003e：\u003c/strong\u003eXuehai Jia and Hao Liu were responsible for data analysis and drafting the manuscript. Kerui Zhang provided technical guidance. Yi Deng, Changyong Shen and Ya Li were responsible for data preprocessing. Litai Ma and Fei Xing were responsible for the experimental design and paper review. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflict of Interest declaration\u003c/strong\u003e: The authors declare that they have no affiliations with or involvement in any organization or entity with any financial interest in the subject matter or materials discussed in this manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics Statement\u003c/strong\u003e\u003cstrong\u003e：\u003c/strong\u003eThe study was conducted in accordance with the Declaration of Helsinki. The data of human received ethical approval from the Ethics Review Committee of West China Hospital of Sichuan University (Approval Number: 2023 Review No. 2427). Informed consent was obtained from all subjects whose CT images were used in this study, and all data were anonymized to protect patient p\u003cstrong\u003erivacy.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding Information\u003c/strong\u003e\u003cstrong\u003e：\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eThe present study received no financial support from either public, commercial, or not-for-profit sources.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication:\u0026nbsp;\u003c/strong\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials:\u0026nbsp;\u003c/strong\u003eThe datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements:\u0026nbsp;\u003c/strong\u003eThe authors thank the orthopedists, Fuguo Huang, Yue Fang and Shiqiang Cen, for their help in this study.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eS. Str\u0026oslash;m R\u0026ouml;nnquist, B. Viberg, M.T. Kristensen, H. Palm, J.-E.B. Jensen, C.F. Madsen, K.E. \u0026Aring;kesson, S. Overgaard, C. Rogmark, Frailty and osteoporosis in patients with hip fractures under the age of 60\u0026mdash;a prospective cohort of 218 individuals, Osteoporosis International 33(5) (2022) 1037-1055.\u003c/li\u003e\n\u003cli\u003eY.W. Zhang, P.P. Lu, Y.J. Li, G.C. Dai, M.H. Chen, Y.K. Zhao, M.M. Cao, Y.F. Rui, Prevalence, Characteristics, and Associated Risk Factors of the Elderly with Hip Fractures: A Cross-Sectional Analysis of NHANES 2005-2010, Clin Interv Aging 16 (2021) 177-185.\u003c/li\u003e\n\u003cli\u003eY.W. Zhang, M.M. Cao, Y.J. Li, G.C. Dai, P.P. Lu, M. Zhang, L.Y. Bai, X.X. Chen, L. Shi, C. Zhang, Y.F. Rui, Dietary Protein Intake in Relation to the Risk of Osteoporosis in Middle-Aged and Older Individuals: A Cross-Sectional Study, J Nutr Health Aging 26(3) (2022) 252-258.\u003c/li\u003e\n\u003cli\u003eY.W. Zhang, P.P. Lu, Y.J. Li, G.C. Dai, M.M. Cao, T. Xie, C. Zhang, L. Shi, Y.F. Rui, Low dietary choline intake is associated with the risk of osteoporosis in elderly individuals: a population-based study, Food Funct 12(14) (2021) 6442-6451.\u003c/li\u003e\n\u003cli\u003eB.L.L. Wong, Y.H. Chan, G.K. O\u0026apos;Neill, D. Murphy, R.A. Merchant, Frailty, length of stay and cost in hip fracture patients, Osteoporos Int 34(1) (2023) 59-68.\u003c/li\u003e\n\u003cli\u003eL. Li, J. Zhen, L. Huang, J. Zhou, L. Yao, L. Xu, W. Zhang, G. Zhang, Q. Chen, B. Cheng, S. Gong, G. Cai, R. Jiang, J. Yan, The risk factors for deep venous thrombosis in critically ill older adult patients: a subgroup analysis of a prospective, multicenter, observational study, BMC Geriatrics 22(1) (2022) 977.\u003c/li\u003e\n\u003cli\u003eV. Sathiyakumar, S.E. Greenberg, A.A. Jahangir, H.H. Mir, W.T. Obremskey, M.K. Sethi, Impact of type of surgery on deep venous thrombi and pulmonary emboli: a look at twenty seven thousand hip fracture patients, Int Orthop 39(10) (2015) 2017-22.\u003c/li\u003e\n\u003cli\u003eY.C. Lin, S.H. Lee, I.J. Chen, C.H. Chang, C.J. Chang, Y.C. Wang, Y. Chang, P.H. Hsieh, Symptomatic pulmonary embolism following hip fracture: A nationwide study, Thromb Res 172 (2018) 120-127.\u003c/li\u003e\n\u003cli\u003eT. Wang, J. Guo, Y. Long, Y. Yin, Z. Hou, Risk factors for preoperative deep venous thrombosis in hip fracture patients: a meta-analysis, J Orthop Traumatol 23(1) (2022) 19.\u003c/li\u003e\n\u003cli\u003eD. Kumar, A. Elhence, R.K. Rajnish, N. Gahlot, S.K. Yadav, S. Gupta, K. Kunal, P. Khera, Mechanical thromboprophylaxis for hip fractures in elderly patients: a prospective randomized controlled study, Am J Transl Res 15(8) (2023) 5284-5291.\u003c/li\u003e\n\u003cli\u003eJ. Gregson, S. Kaptoge, T. Bolton, L. Pennells, P. Willeit, S. Burgess, S. Bell, M. Sweeting, E.B. Rimm, C. Kabrhel, B. Z\u0026ouml;ller, G. Assmann, V. Gudnason, A.R. Folsom, V. Arndt, A. Fletcher, P.E. Norman, B.G. Nordestgaard, A. Kitamura, B.K. Mahmoodi, P.H. Whincup, M. Knuiman, V. Salomaa, C. Meisinger, W. Koenig, M. Kavousi, H. V\u0026ouml;lzke, J.A. Cooper, T. Ninomiya, E. Casiglia, B. Rodriguez, Y. Ben-Shlomo, J.-P. Despr\u0026eacute;s, L. Simons, E. Barrett-Connor, C. Bj\u0026ouml;rkelund, M. Notdurfter, D. Kromhout, J. Price, S.E. Sutherland, J. Sundstr\u0026ouml;m, J. Kauhanen, J. Gallacher, J.W.J. Beulens, R. Dankner, C. Cooper, S. Giampaoli, J.F. Deen, A. G\u0026oacute;mez de la C\u0026aacute;mara, L.H. Kuller, A. Rosengren, P.J. Svensson, D. Nagel, C.J. Crespo, H. Brenner, J.R. Albertorio-Diaz, R. Atkins, E.J. Brunner, M. Shipley, I. Nj\u0026oslash;lstad, D.A. Lawlor, Y.T. van der Schouw, R.M. Selmer, M. Trevisan, W.M.M. Verschuren, P. Greenland, S. Wassertheil-Smoller, G.D.O. Lowe, A.M. Wood, A.S. Butterworth, S.G. Thompson, J. Danesh, E. Di Angelantonio, T. Meade, f.t.E.R.F. Collaboration, Cardiovascular Risk Factors Associated With Venous Thromboembolism, JAMA Cardiology 4(2) (2019) 163-173.\u003c/li\u003e\n\u003cli\u003eM. Cushman, Epidemiology and risk factors for venous thrombosis, Semin Hematol 44(2) (2007) 62-9.\u003c/li\u003e\n\u003cli\u003eF.A. Anderson, F.A. Spencer, Risk Factors for Venous Thromboembolism, Circulation 107(23_suppl_1) (2003) I-9-I-16.\u003c/li\u003e\n\u003cli\u003eB.-H. Wang, Y.-D. Sun, X.-C. Fan, B.-F. Zhang, The admission pH is a risk factor of preoperative deep vein thrombosis in geriatric hip fracture: a retrospective cohort study, Scientific Reports 13(1) (2023) 18392.\u003c/li\u003e\n\u003cli\u003eJ. Jiang, F. Xing, R. Luo, Z. Chen, H. Liu, Z. Xiang, X. Duan, Risk factors and prediction model of nomogram for preoperative calf muscle vein thrombosis in geriatric hip fracture patients, Frontiers in Medicine 10 (2023).\u003c/li\u003e\n\u003cli\u003eL. Langsetmo, J.T. Schousboe, B.C. Taylor, J.A. Cauley, H.A. Fink, P.M. Cawthon, D.M. Kado, K.E. Ensrud, Advantages and Disadvantages of Random Forest Models for Prediction of Hip Fracture Risk Versus Mortality Risk in the Oldest Old, JBMR Plus 7(8) (2023) e10757.\u003c/li\u003e\n\u003cli\u003eJ. Wu, J. Qiu, E. Xie, W. Jiang, R. Zhao, J. Qiu, M.A. Zafar, Y. Huang, C. Yu, Predicting in-hospital rupture of type A aortic dissection using Random Forest, Journal of Thoracic Disease 11(11) (2019) 4634-4646.\u003c/li\u003e\n\u003cli\u003eX. Han, X. Zheng, Y. Wang, X. Sun, Y. Xiao, Y. Tang, W. Qin, Random forest can accurately predict the development of end-stage renal disease in immunoglobulin a nephropathy patients, Ann Transl Med 7(11) (2019) 234.\u003c/li\u003e\n\u003cli\u003eP.C. Kruger, J.W. Eikelboom, J.D. Douketis, G.J. Hankey, Deep vein thrombosis: update on diagnosis and management, Med J Aust 210(11) (2019) 516-524.\u003c/li\u003e\n\u003cli\u003eL. Duffett, Deep Venous Thrombosis, Annals of Internal Medicine 175(9) (2022) ITC129-ITC144.\u003c/li\u003e\n\u003cli\u003eL.-K. Liu, W.-J. Lee, L.-Y. Chen, A.-C. Hwang, M.-H. Lin, L.-N. Peng, L.-K. Chen, Association between Frailty, Osteoporosis, Falls and Hip Fractures among Community-Dwelling People Aged 50 Years and Older in Taiwan: Results from I-Lan Longitudinal Aging Study, PLOS ONE 10(9) (2015) e0136968.\u003c/li\u003e\n\u003cli\u003eJ. Stone, P. Hangge, H. Albadawi, A. Wallace, F. Shamoun, M.G. Knuttien, S. Naidu, R. Oklu, Deep vein thrombosis: pathogenesis, diagnosis, and medical management, Cardiovascular Diagnosis and Therapy (2017) S276-S284.\u003c/li\u003e\n\u003cli\u003eY. Zhou, H. Chen, Y. Zhang, C. Yang, X. Yi, S. Liu, Y. Zeng, Computational Analysis of the Related Factors of Deep Vein Thrombosis (DVT) Formation in Patients Undergoing Hip Fracture Surgery, Evidence-Based Complementary and Alternative Medicine 2022 (2022) 1127095.\u003c/li\u003e\n\u003cli\u003eE. Kavalci, A. Hartshorn, Improving clinical trial design using interpretable machine learning based prediction of early trial termination, Sci Rep 13(1) (2023) 121.\u003c/li\u003e\n\u003cli\u003eM. Toma, O.C. Wei, Predictive Modeling in Medicine, Encyclopedia 3(2) (2023) 590-601.\u003c/li\u003e\n\u003cli\u003eK.-M. Kim, J.-H. Kim, H.-S. Rhee, B.-Y. Youn, Development of a prediction model for the depression level of the elderly in low-income households: using decision trees, logistic regression, neural networks, and random forest, Scientific Reports 13(1) (2023) 11473.\u003c/li\u003e\n\u003cli\u003eF. Xing, R. Luo, M. Liu, Z. Zhou, Z. Xiang, X. Duan, A New Random Forest Algorithm-Based Prediction Model of Post-operative Mortality in Geriatric Patients With Hip Fractures, Frontiers in Medicine 9 (2022).\u003c/li\u003e\n\u003cli\u003eX. Liu, X. Jiao, X. Gong, Q. Nie, Y. Li, G. Zhen, M. Cheng, J. He, Y. Yuan, Y. Yang, Prevalence, Risk Factor and Clinical Characteristics of Venous Thrombus Embolism in Patients with Acute Exacerbation of COPD: A Prospective Multicenter Study, Int J Chron Obstruct Pulmon Dis 18 (2023) 907-917.\u003c/li\u003e\n\u003cli\u003eJ. Hou, W. Wang, H. Cai, J. Chen, B. Chen, Z. Shen, Y. Tang, J. Li, S. Liu, Y. Mei, J. Wang, S. Lu, Patients With Right Lower Extremity Deep Vein Thrombosis Have a Higher Risk of Symptomatic Pulmonary Embolism: A Retrospective Study of 1585 Patients, Annals of Vascular Surgery 81 (2022) 240-248.\u003c/li\u003e\n\u003cli\u003eJ.R. Byrnes, A.S. Wolberg, Red blood cells in thrombosis, Blood 130(16) (2017) 1795-1799.\u003c/li\u003e\n\u003cli\u003eT. Watson, E. Shantsila, G.Y. Lip, Mechanisms of thrombogenesis in atrial fibrillation: Virchow\u0026apos;s triad revisited, Lancet 373(9658) (2009) 155-66.\u003c/li\u003e\n\u003cli\u003eG. Alkarithi, C. Duval, Y. Shi, F.L. Macrae, R.A.S. Ariens, Thrombus Structural Composition in Cardiovascular Disease, Arterioscler Thromb Vasc Biol 41(9) (2021) 2370-2383.\u003c/li\u003e\n\u003cli\u003eW. Yang, Q. Wei, H. Wang, K. Ding, M. Li, C. Li, C. Liang, Y. Zhu, W. Chen, Preoperative incidence and risk factors of deep venous thrombosis in patients with isolated femoral shaft fracture, BMC Surgery 22(1) (2022) 83.\u003c/li\u003e\n\u003cli\u003eT. Taoka, T. Ohmori, T. Kanazawa, K. Toda, T. Ishihara, Y. Ito, Delayed surgery after hip fracture affects the incidence of venous thromboembolism, Journal of Orthopaedic Surgery and Research 18(1) (2023) 630.\u003c/li\u003e\n\u003cli\u003eJ. Cheng, Z.b. Fu, J. Zhu, L. Zhou, W. Song, The predictive value of plasminogen activator inhibitor-1, fibrinogen, and D-dimer for deep venous thrombosis following surgery for traumatic lower limb fracture, Annals of palliative medicine (2020).\u003c/li\u003e\n\u003cli\u003eT. Wang, J. Guo, Y. Long, Y. Yin, Z. Hou, Risk factors for preoperative deep venous thrombosis in hip fracture patients: a meta-analysis, Journal of Orthopaedics and Traumatology 23(1) (2022) 19.\u003c/li\u003e\n\u003cli\u003eY. Bai, M. Shi, X. Yang, W. Zhang, R. Yang, X. Wei, X. Wei, L. Duan, C. Wang, R. Mi, H.A. Waqas Ahmed, L. Huo, Y. Chen, F. Xu, D. Wu, K. Sun, The value of FDP/FIB and D-dimer/FIB ratios in predicting high-risk APL-related thrombosis, Leukemia Research 79 (2019) 34-37.\u003c/li\u003e\n\u003cli\u003eM. Pang, F. Zhao, P. Yu, X. Zhang, H. Xiao, W. Qiang, H. Zhu, L. Zhao, The significance of coagulation and fibrinolysis-related parameters in predicting postoperative venous thrombosis in patients with breast cancer, Gland Surg 10(4) (2021) 1439-1446.\u003c/li\u003e\n\u003cli\u003eS. Yang, T. Wang, Y. Long, J. Guo, Z. Hou, Risk Factors of Deep Vein Thrombosis in Adults with Acute Compartment Syndrome Following Lower Extremity Fractures, Clinical and Applied Thrombosis/Hemostasis 29 (2023) 10760296231165053.\u003c/li\u003e\n\u003cli\u003eY. Sun, J. Deng, Y. Ding, S. Luo, S. Li, Y. Guan, X. Cao, X. Hao, Y. Hu, Serum albumin, genetic susceptibility, and risk of venous thromboembolism, Research and Practice in Thrombosis and Haemostasis 8(5) (2024).\u003c/li\u003e\n\u003cli\u003eY.-L. Wu, D. Zhang, K.-Y. Zhang, T. Yan, W.-S. Qiang, T. Zhang, B.-F. Zhang, The association between admission serum albumin and preoperative deep venous thrombosis in geriatrics hip fracture: a retrospective study of 1819 patients with age\u0026thinsp;\u0026ge;\u0026thinsp;65 years, BMC Musculoskeletal Disorders 24(1) (2023) 672.\u003c/li\u003e\n\u003cli\u003eW. Yao, K. Zhang, Q. Lv, Z. Deng, W. Ding, D-dimer-albumin ratio (DAR) as a new biomarker for predicting preoperative deep vein thrombosis after geriatric hip fracture patients, Journal of Orthopaedic Surgery and Research 18(1) (2023) 645.\u003c/li\u003e\n\u003cli\u003eN. Cay, A. Ipek, M. Gumus, Z. Birkan, E. Ozmen, Platelet activity indices in patients with deep vein thrombosis, Clin Appl Thromb Hemost 18(2) (2012) 206-10.\u003c/li\u003e\n\u003cli\u003eL. Zhang, M. He, W. Jia, W. Xie, Y. Song, H. Wang, J. Peng, Y. Li, Z. Wang, Z. Lin, Analysis of high-risk factors for preoperative DVT in elderly patients with simple hip fractures and construction of a nomogram prediction model, BMC Musculoskeletal Disorders 23(1) (2022) 441.\u003c/li\u003e\n\u003cli\u003eS. Niu, J. Li, Y. Zhao, D. Ding, G. Jiang, Z. Song, Preoperative deep venous thrombosis (DVT) after femoral neck fracture in the elderly, the incidence, timing, location and related risk factors, BMC Musculoskeletal Disorders 22(1) (2021) 264.\u003c/li\u003e\n\u003cli\u003eZ. Wang, X. Chen, J. Wu, Q. Zhou, H. Liu, Y. Wu, S. Liu, Y. Liu, Low Mean Platelet Volume is Associated with Deep Vein Thrombosis in Older Patients with Hip Fracture, Clinical and Applied Thrombosis/Hemostasis 28 (2022) 10760296221078837.\u003c/li\u003e\n\u003cli\u003eM.J. Huang, R.B. Wei, Y. Wang, T.Y. Su, P. Di, Q.P. Li, X. Yang, P. Li, X.M. Chen, Blood coagulation system in patients with chronic kidney disease: a prospective observational study, BMJ Open 7(5) (2017) e014294.\u003c/li\u003e\n\u003cli\u003eJ. Haltout, A. Awada, M. Paesmans, M. Moreau, J. Klastersky, G. Machiels, M. Ignatiadis, N. Kotecki, Predictive factors for cancer-associated thrombosis in a large retrospective single-center study, Supportive Care in Cancer 27(4) (2019) 1163-1170.\u003c/li\u003e\n\u003cli\u003eZ. Wang, J. Xiao, Z. Zhang, X. Qiu, Y. Chen, Chronic kidney disease can increase the risk of preoperative deep vein thrombosis in middle-aged and elderly patients with hip fractures, Clin Interv Aging 13 (2018) 1669-1674.\u003c/li\u003e\n\u003cli\u003eQ. Li, B. Dai, Y. Yao, K. Song, D. Chen, Q. Jiang, Chronic Kidney Dysfunction Can Increase the Risk of Deep Vein Thrombosis after Total Hip and Knee Arthroplasty, BioMed Research International 2017 (2017) 8260487.\u003c/li\u003e\n\u003cli\u003eD. Gruson, A. Buglioni, J.C. Burnett, Jr., PTH: Potential role in management of heart failure, Clin Chim Acta 433 (2014) 290-6.\u003c/li\u003e\n\u003cli\u003eG. Olmos, P. Martinez-Miguel, E. Alcalde-Estevez, D. Medrano, P. Sosa, L. Rodriguez-Manas, M. Naves-Diaz, D. Rodriguez-Puyol, M.P. Ruiz-Torres, S. Lopez-Ongil, Hyperphosphatemia induces senescence in human endothelial cells by increasing endothelin-1 production, Aging Cell 16(6) (2017) 1300-1312.\u003c/li\u003e\n\u003cli\u003eD.-X. Lu, K. Zhang, T. Ma, M. Li, Z. Li, Y.-B. Xu, C.-F. Wang, C. Ren, B.-F. Zhang, The Association between Admission Serum Phosphorus and Preoperative Deep Venous Thrombosis in Geriatric Hip Fracture: A Retrospective Study, Diagnostics 13(3) (2023) 545.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"machine learning, random forest, HF, deep vein thrombosis, prediction model","lastPublishedDoi":"10.21203/rs.3.rs-6339181/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6339181/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground: \u003c/strong\u003eDeep vein thrombosis (DVT) poses a common and critical risk for mortality in elderly hip fracture (HF) patients. Venous angiography and ultrasound examinations serve as crucial diagnostic tools but pose challenges in cases with prevalent complications. The extensive training period for technical personnel, coupled with the rapid advancements in machine learning, prompts our research to harness the potential of the random forest algorithm. Our aim is to construct a predictive model that evaluates the risk of thrombosis formation in elderly hip fracture patients upon admission.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods:\u003c/strong\u003eWe conducted a retrospective evaluation of 448 elderly HF patients who received surgical treatment between May 2021 and November 2023. The study cohort was partitioned into training and test datasets, maintaining a 70:30 ratio. Leveraging the Random Forest algorithm, we developed a streamlined predictive model.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults:\u003c/strong\u003eEleven important variables, namely ALB, A/G, PLT, Fib, D-dimer, CREA, Pi, GFR, FDP, fracture onset to surgery time, and cardiopulmonary diseases, were screened based on Random Forest features. In the training set, AUC, the 95% Confidence Interval (CI), Sensitivity, Specificity, Precision, Accuracy, and Balanced Accuracy stand at 1.000, (0.9883, 1), 1.000, 1.000, 1.000, 1.000, and 1.000, respectively. For the test set, AUC, the 95% CI, Sensitivity, Specificity, Precision, Accuracy, and Balanced Accuracy are 0.899, (0.7875, 0.9124), 0.6061, 0.9406, 0.7692, 0.8582, and 0.7733, respectively.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions:\u003c/strong\u003eA random forest prediction model was developed to anticipate the occurrence of preoperative lower extremity DVT in elderly HF patients. This model demonstrated superior accuracy compared to the logistic regression model. Key preoperative laboratory test indicators proved valuable as variables in the prediction process.\u003c/p\u003e","manuscriptTitle":"The Role of Preoperative Laboratory Test Indicators in Predicting Thrombosis Risk in Elderly Hip Fracture Patients: A Random Forest Approach","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-29 15:46:20","doi":"10.21203/rs.3.rs-6339181/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"9653aef8-25a3-41ce-a037-1f155bcbe430","owner":[],"postedDate":"July 29th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":52069309,"name":"Biological sciences/Physiology/Bone"},{"id":52069310,"name":"Health sciences/Medical research/Outcomes research"}],"tags":[],"updatedAt":"2025-12-22T12:52:15+00:00","versionOfRecord":[],"versionCreatedAt":"2025-07-29 15:46:20","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6339181","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6339181","identity":"rs-6339181","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}