Machine Learning and Transformer models for Prediction of Postoperative Pneumonia Risk in Patients with Lower Limb Fractures

Machine Learning and Transformer models for Prediction of Postoperative Pneumonia Risk in Patients with Lower Limb Fractures | 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 Machine Learning and Transformer models for Prediction of Postoperative Pneumonia Risk in Patients with Lower Limb Fractures Yiqun Chen, Mingxuan Ma, Dandan Qu, Chunxiang Xu This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6056672/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 01 Jul, 2025 Read the published version in Scientific Reports → Version 1 posted 5 You are reading this latest preprint version Abstract Background: Postoperative pneumonia, a prevalent complication arising from lower limb fracture surgery, can significantly prolong hospitalization periods and elevate mortality rates. Consequently, early prevention and identification of this condition are crucial in improving patient prognosis. Methods: In this study, clinical indicators pertaining to postoperative pneumonia in patients with lower limb fractures at Nantong University Hospital, spanning the years 2016 to 2023, were subjected to a analysis. The patients who encountered postoperative pneumonia subsequent to their lower limb fracture surgeries during hospitalization were categorized as the case group, whereas those who did not develop such a complication served as the control group. To forecast the likelihood of postoperative pneumonia occurrence, both machine learning and deep learning algorithms were employed. Results: The study identified Age, Gender, Fracture type, Venous thromboembolism (VTE), Hypertension, Chronic obstructive pulmonary disease (COPD), Cancer, Atrial fibrillation, Cerebrovascular disease, Hypoalbuminemia, Free fatty acid, Albumin, Albumin to globulin ratio, Calcium, Fibrinogen, D-dimer, Alcohol, Surgical grade and C-reactive protein as significant predictors of postoperative pneumonia. XGBoost and Transformer models have better performance (AUC 0.866 VS 0.966 , F1 0.807 VS 0.889), and both models have better substantial prediction ability for the occurrence of postoperative pneumonia. Conclusion: In conclusion, XGBoost and Transformer models serve as potential tools for the prevention and treatment of postoperative pneumonia in patients with lower-extremity fractures. By adopting appropriate health management practices, the risk of developing postoperative pneumonia in this patient population may be reduced. Health sciences/Medical research/Translational research Biological sciences/Computational biology and bioinformatics/Machine learning Postoperative pneumonia risk factors lower extremity fracture machine learning model Transformer Figures Figure 1 Figure 2 Figure 3 Introduction Postoperative pneumonia is a prevalent complication following lower extremity fracture surgery, with an incidence ranging from 5.1–14.9%[1]. The increasing number of such surgeries over the past few decades has underscored the significance of this complication within the healthcare and social support systems. Epidemiological studies indicate that postoperative comorbid pneumonia in patients with lower extremity fractures can lead to a mortality rate of 27–43%, extend hospitalization by 56%[2, 3], and escalate the risk of readmission by a factor of eight. These adverse outcomes are often attributed to factors such as advanced age, comorbidities, surgical trauma, prolonged limb immobilization, and diminished immune and pulmonary function[4–6]. Consequently, identifying modifiable risk factors for the development of postoperative pneumonia in this patient population is crucial for improving patient outcomes[7]. Accurately identifying high-risk patients necessitates the determination of significant predictors of postoperative pneumonia, a complex outcome influenced by a myriad of interacting factors. The occurrence of postoperative pneumonia exhibits considerable diversity and instability, with its risk factors continually under investigation. Prior research has extensively delved into potential risk factors for postoperative pneumonia complicating lower extremity fractures, encompassing variables such as gender, age, anemia, surgical duration, hospital stay length, and select laboratory biomarkers[6]. These studies have enhanced our comprehension of factors that elevate the risk of postoperative pneumonia. However, they are constrained by limitations like small sample sizes, incomplete consideration of risk factors, and inclusion of select, older populations, thereby impeding the generalizability of their findings and conclusions. The disease prediction model, a quantitative instrument for assessing disease risk, has emerged as a crucial auxiliary in identifying high-risk populations during clinical diagnosis, treatment, and nursing care. The essence of achieving precise disease prediction lies in the thorough examination of the interplay between disease-related factors. Nevertheless, the variables incorporated in existing postoperative lung infection risk models for lower extremity fractures, as documented in the literature, are largely based on personal experience or subjective consultations, encompassing elements such as smoking history and the number of preoperative comorbidities[8, 9]. Additionally, these models may incorporate intricate scoring tools, such as functional status assessments, which can pose challenges for clinical caregivers in terms of practical application. Consequently, we conducted a retrospective analysis of objective risk factors for postoperative pneumonia following lower extremity fractures. Leveraging machine learning and deep learning algorithms, we developed an advanced and intelligent tool to aid clinical caregivers in the early prevention of postoperative pneumonia. Methods Study Population This retrospective cohort study examined two distinct patient groups (pneumonia and non-pneumonia) who had undergone surgery for lower limb fractures at the Affiliated Hospital of Nantong University. The study population encompassed individuals with fractures in the hip, femur, tibia, and fibula. The medical records of these patients were retrospectively analyzed over an eight-year period, spanning from January 2015 to January 2023, yielding a total of 4,424 cases for evaluation. Randomly divided into training dataset (n = 3539) and test set (n = 1065) according to 8:2 data. Inclusion and Exclusion Inclusion criteria encompassed individuals diagnosed with lower limb fractures, encompassing tibial, femoral, and pelvic fractures, based on a comprehensive assessment of clinical symptoms, signs, and imaging findings. Exclusion criteria included those with a preexisting diagnosis of pneumonia prior to surgery, pathological fractures, or combined fractures in other anatomical regions, such as ribs, upper limb bones, or vertebrae. Additionally, patients who expired for any reason during hospitalization were excluded from the study. To assess the development of postoperative pneumonia, researchers thoroughly reviewed medical records from the day subsequent to surgery until hospital discharge. A rigorous data entry process involving double entry and cross-checking was implemented to minimize potential errors. Any discrepancies encountered during this process were resolved through collaborative discussion and consensus among the research team. Definition of Pneumonia The diagnosis of pneumonia in this study adhered to the guidelines established by the American Thoracic Society for healthcare-associated pneumonia, augmented by relevant online resources[10]. The diagnostic criteria encompassed the presence of novel, progressive, or persistent respiratory symptoms, inclusive of coughing and the production of purulent secretions. Additionally, patients exhibited either fever (body temperature exceeding 38°C) or hypothermia (body temperature falling below 36°C). Confirmation of lung consolidation or moist rales through physical examination, along with laboratory findings indicative of leukocytosis (white blood cell count exceeding 10×109/L) or leukopenia (white blood cell count falling below 4×109/L), further strengthened the diagnosis. Finally, positive results from blood cultures or sputum samples provided definitive evidence of pneumonia. Feature Selection We identified potential predictors of pneumonia formation by selecting factors that significantly impact its development. This was accomplished through a comprehensive literature review and clinical expertise. The resulting list of predictors included age, gender,Fracture type, VTE, diabetes, hypertension, COPD, cancer, heart disease, cerebrovascular disease, atrial fibrillation, hypoalbuminemia, total cholesterol, Free fatty acid, Albumin, Albumin to globulin ratio, calcium, potassium, Platelet, red blood cell, leukocyte, Fibrinogen, D-dimer, Length of stay, smoking and alcohol consumption history, operation time, blood group, Surgical grade and C-reactive protein levels. Operation time was dichotomized based on a threshold of 3 hours. To assess the correlation between features with high repeatability, Spearman's rank correlation coefficient was utilized (as depicted in Fig. 1 , which shows the Spearman correlation of each feature). In cases where the correlation coefficient exceeded 0.9 between any two features, only one of the highly correlated features was retained to avoid redundancy. To maximize the representation of features while minimizing redundancy, a greedy recursive deletion strategy was employed for feature filtering. Specifically, the feature exhibiting the greatest redundancy within the current set was iteratively removed. Ultimately, this process resulted in the retention of 23 features. The Least Absolute Shrinkage and Selection Operator (LASSO) regression model was utilized on the discovery dataset to construct a radiomics signature. By adjusting the regularization parameter λ, LASSO shrinks all regression coefficients towards zero, effectively eliminating the coefficients of irrelevant features by setting them exactly to zero. To identify an optimal λ, we employed 10-fold cross-validation with a minimum error criterion, ensuring that the final λ value minimized the cross-validation error. The retained features with nonzero coefficients were then incorporated into a regression model, forming the clinical signature. Subsequently, we derived a clinical score for each patient by linearly combining the retained features, with each feature weighted by its corresponding model coefficient. This analysis was facilitated by the Python scikit-learn package for LASSO regression modeling. SMOTE-ENN Resampling In order to solve the problem of class imbalance in the data set, we implemented the SMOTE-ENN hybrid resampling technique in the training set before the model development. After resampling, the ratio: adjusted minority: majority is 1:1.2 (1:14.5 than the original). At the same time strictly ensure the independence of test set data. This method combines the synthetic minority oversampling technique (SMOTE) for minority class expansion and the edited nearest neighbor (ENN) for most class noise reduction, effectively balancing the class distribution while improving the clarity of decision boundaries. Model building In the process of constructing predictive models using machine learning algorithms, clinical variables that have undergone screening are incorporated as parameter features. Eight machine learning algorithms[11], including Support Vector Machines (SVM), K-Nearest Neighbors (KNN), Multilayer Perceptron (MLP), ExtraTrees (ET), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Adaptive Boosting (AdaBoost), are employed to establish these models. To ensure the stability and reproducibility of model performance, 10-fold cross-validation resampling is utilized. In contrast, in the Transformer model, all features are incorporated for deep learning training without prior screening to select the optimal model. Statistical Analysis Statistical analysis was performed using SPSS 26.0 (IBM, Armonk, NY, USA). The incidence of postoperative pneumonia was determined by dividing the number of patients who developed pneumonia during hospitalization by the total number of patients included in the study. Continuous variables with normal distribution were expressed as mean ± standard deviation, while those with non-normal distribution were expressed as medians (interquartile range), following confirmation of normality using the Shapiro-Wilk test. Subsequently, either a t-test or Whitney-U test was employed as appropriate. Categorical variables were expressed as numbers with percentages, and group differences were analyzed using Chi-square or Fisher's exact test. Statistical significance was set at P < 0.05. To assess the predictive effectiveness of the column-line diagrams, ROC curves were constructed. The clinical utility of the predictive models was evaluated using DCA decision curves, and calibration curves were generated to assess the calibration efficiency of the column-line diagrams. Results Clinical Characteristics Table 1 summarizes the patient characteristics in the training set. We reviewed the medical records of 3311 patients without pneumonia and 228 patients with pneumonia, yielding a pneumonia incidence rate of 6.44%. Statistical significance was observed between the no pneumonia and pneumonia groups for age (P < 0.001), Gender (P = 0.803), fracture type (P < 0.001), VTE (P = 0.766), Diabetes (P = 0.0948) Hypertension (P = 0.96), COPD (P < 0.001), Cancer (P < 0.001), Heart disease (P = 0.517), Atrial fibrillation (P < 0.001), Cerebrovascular disease (P < 0.001), Nephropathy (P = 0.934), Hypoalbuminemia (P < 0.001), Total cholesterol (P = 0.971), Free fatty acid (P = 0.0146), Albumin(P < 0.001), Albumin to globulin ratio (P < 0.001), calcium (P < 0.001), potassium (P = 0.969), Platelet (P = 0.13), Red blood cells (P = 0.983), Leukocyte (P = 0.971), Fibrinogen(P < 0.001), D-dimer (P = 0.001), Length of stay (P = 0.781), Operation time (P = 0.382), Alcohol disease (P = 0.190), smoking disease (P = 0.162), CRP (P < 0.001), Blood group (P = 0.001) and Surgical grade (P = 0.395) . To establish the model, we employed a least absolute shrinkage and selection operator (LASSO) logistic regression model, selecting only nonzero coefficients. The coefficients and mean squared error (MSE) from 10-fold cross-validation are depicted in Fig. 1 . Nineteen non-zero coefficients were identified, including Age, Gender, Fracture type, VTE, Hypertension, COPD, Cancer, Atrial fibrillation, Cerebrovascular disease, Hypoalbuminemia, Free fatty acid, Albumin, Albumin to globulin ratio, Calcium, Fibrinogen, D-dimer, Alcohol, Surgical grade and CRP. Comparison of machine learning model and deep learning model Eight machine learning models and a deep learning model were used to evaluate the performance of the test set. As can be seen from Table 2 , the AUC of the SVM model is 0.764, F1-score is 0.666, the AUC of the KNN model is 0.812, F1-score is 0.755, the AUC of the random forest model is 0.834, F1-score is 0.760, the AUC of the ExtraTrees model is 0.843, F1-score is 0.768, the AUC of the LightGBM model is 0.869, F1-score is 0.778, the AUC of the AdaBoost model is 0.852, F1-score is 0.768, the AUC of the MLP model is 0.859, F1-score is 0.763, and the AUC of the XGBoost model is 0.866, F1-score is 0.807, the AUC of the Transformer model is 0.946, and the F1-score is 0.889. The XGBoost model and the Transformer model have the best performance among many models. Figure 2 shows the ROC curve, decision curve and calibration curve of the two optimal models in the test set. It can also be seen from Fig. 2 A that both the XGBoost and transformer models show superior performance, with AUC values of 0.866 and 0.946, respectively. From Fig. 2 B, the transformer model dominates in the same cohort study. Moreover, the decision curve area of the transformer model is the largest in Fig. 2 C. In addition, from Fig. 2 D, the oblique of the transformer model calibration curve is close to 1, indicating that the risk of postoperative pneumonia predicted by the transformer model is in good agreement with the actual risk. Model Visualization We further plotted a summary plot of SHAP values to interpret the XGBoost model results (Fig. 3 A). For each feature, a point corre sponds to a patient, and the position of the point on the x-axis indicates the effect of the feature on the model output for that particular patient. In Fig. 3 A, the SHAP values of all indicators are labeled in color (red/blue) to ensure that the contribution direction is clear at a glance. Specifically, age, gender, albumin level, calcium ion level, and history of hypertension occupy a high weight in all characteristics. Among them, albumin and calcium ion levels are negatively correlated, and others are positively correlated. In the individual prediction process, Fig. 3 B, the current patient's age is 82 and the SHAP value is + 0.67. Advanced age is significantly positively correlated with the risk of pneumonia events; albumin level (32.56 g/L, lower than the cohort average) contributes to the SHAP value of + 0.14, indicating that hypoalbuminemia increases the risk of pneumonia; calcium ion level is 2.026 (lower than the cohort average), and the SHAP value is + 0.16, indicating that low calcium also enhances the risk of events. The corresponding SHAP values of other indicators such as history of hypertension, cerebrovascular disease, high D-dimer, and high CRP all indicate that the risk of events is increased. For gender, women tend to have a lower risk of pneumonia after fracture surgery. Table 1 Clinical characters in the training set. Train No pneumonia (N = 3311) Pneumonia (N = 228) P -value Age < 0.001 Median (range) 60.0 [2.00, 102] 75.0 [12.0, 96.0] Gender 0.803 Male 1765 (53.3%) 129 (56.6%) Female 1546 (46.7%) 99.0 (43.4%) Fracture type < 0.001 Hip 147 (4.4%) 6.00 (2.6%) Femur 1996 (60.3%) 167 (73.2%) Tibial 1005 (30.4%) 24.0 (10.5%) Multiple 163 (4.9%) 31.0 (13.6%) VTE 0.766 No 3157 (95.3%) 215 (94.3%) Yes 154 (4.7%) 13.0 (5.7%) Diabetes 0.0948 No 2888 (87.2%) 192 (84.2%) Yes 423 (12.8%) 36.0 (15.8%) Hypertension 0.96 No 2412 (72.8%) 162 (71.1%) Yes 899 (27.2%) 66.0 (28.9%) COPD < 0.001 No 3221 (97.3%) 200 (87.7%) Yes 90.0 (2.7%) 28.0 (12.3%) Cancer < 0.001 No 3118 (94.2%) 200 (87.7%) Yes 193 (5.8%) 28.0 (12.3%) Heart disease 0.517 No 3156 (95.3%) 214 (93.9%) Yes 155 (4.7%) 14.0 (6.1%) Atrial fibrillation < 0.001 No 3168 (95.7%) 208 (91.2%) Yes 143 (4.3%) 20.0 (8.8%) Cerebrovascular disease < 0.001 No 2943 (88.9%) 188 (82.5%) Yes 368 (11.1%) 40.0 (17.5%) Nephropathy 0.934 No 3248 (98.1%) 222 (97.4%) Yes 63.0 (1.9%) 6.00 (2.6%) Hypoalbuminemia < 0.001 No 3177 (96.0%) 209 (91.7%) Yes 134 (4.0%) 19.0 (8.3%) Total cholesterol 0.971 Median (range) 3.74 [0.700, 7.91] 3.74 [1.92, 7.14] Free fatty acid 0.0146 Median (range) 0.510 [0.0100, 4.09] 0.550 [0.0400, 1.94] Albumin < 0.001 Median (range) 37.0 [13.3, 54.3] 34.3 [18.9, 48.6] Albumin to globulin ratio < 0.001 Median (range) 1.42 [0.370, 3.25] 1.33 [0.670, 2.20] calcium < 0.001 Median (range) 2.24 [1.70, 3.04] 2.17 [0.930, 2.63] potassium 0.969 Median (range) 3.88 [2.00, 7.43] 3.91 [2.41, 8.25] Platelet 0.13 Median (range) 186 [13.0, 1470] 167 [23.0, 1220] Red blood cell 0.983 Median (range) 3.83 [0, 2880] 3.62 [1.25, 392] Leukocyte 0.971 Median (range) 8.04 [1.34, 1000] 8.69 [2.54, 38.4] Fibrinogen < 0.001 Median (range) 3.33 [0.300, 11.5] 3.64 [0.950, 7.83] D dimer 0.00106 Median (range) 5.88 [0.100, 103] 7.81 [0.650, 153] Length of stay 0.781 Median (range) 12.0 [1.00, 93.0] 13.0 [1.00, 59.0] Operation time 0.382 ≤ 3h 3030 (91.5%) 206 (90.4%) >3h 281 (8.5%) 22.0 (9.6%) Alcohol 0.19 No 3225 (97.4%) 217 (95.2%) Yes 86.0 (2.6%) 11.0 (4.8%) Smoking 0.162 No 3211 (97.0%) 216 (94.7%) Yes 100 (3.0%) 12.0 (5.3%) CRP < 0.001 Median (range) 41.4 [0.250, 333] 51.6 [0.500, 200] Blood group 0.00162 A 1059 (32.0%) 63.0 (27.6%) B 1296 (39.1%) 101 (44.3%) AB 724 (21.9%) 45.0 (19.7%) O 232 (7.0%) 19.0 (8.3%) Surgical grade 0.0395 I 21.0 (0.6%) 0 (0%) II 119 (3.6%) 3.00 (1.3%) III 2513 (75.9%) 169 (74.1%) IV 658 (19.9%) 56.0 (24.6%) Table 2 Performance Demonstration of Machine Learning Models and Transformer Model. Model ACC AUC 95%CI Sensitivity Specificity Precision Recall F1 Task SVM 0.752 0.820 0.8097–0.8310 0.708 0.788 0.731 0.708 0.719 train SVM 0.705 0.767 0.7428–0.7904 0.655 0.745 0.676 0.655 0.666 test KNN 0.774 0.829 0.8015–0.8557 0.684 0.872 0.854 0.684 0.760 train KNN 0.760 0.812 0.8004–0.8241 0.713 0.810 0.802 0.713 0.755 test RandomForest 0.769 0.837 0.8107–0.8626 0.734 0.809 0.807 0.734 0.769 train RandomForest 0.759 0.834 0.8231–0.8445 0.733 0.787 0.788 0.733 0.760 test ExtraTrees 0.794 0.856 0.8313–0.8799 0.846 0.738 0.779 0.846 0.811 train ExtraTrees 0.764 0.843 0.8321–0.8529 0.753 0.776 0.783 0.753 0.768 test MLP 0.797 0.875 0.8572–0.8919 0.825 0.774 0.748 0.825 0.784 train MLP 0.783 0.859 0.8496–0.8683 0.781 0.784 0.746 0.781 0.763 test LightGBM 0.797 0.873 0.8561–0.8909 0.831 0.771 0.746 0.831 0.786 train LightGBM 0.798 0.869 0.8598–0.8776 0.684 0.789 0.768 0.789 0.778 test AdaBoost 0.785 0.864 0.8407–0.8872 0.762 0.811 0.815 0.762 0.787 train AdaBoost 0.771 0.852 0.8422–0.8622 0.733 0.811 0.807 0.733 0.768 test XGBoost 0.800 0.876 0.8668–0.8851 0.846 0.744 0.781 0.846 0.812 train XGBoost 0.786 0.866 0.8432–0.8894 0.855 0.712 0.764 0.855 0.807 test Transformer 0.908 0.964 0.9596–0.9687 0.919 0.895 0.904 0.919 0.912 train Transformer 0.887 0.946 0.9308–0.9609 0.870 0.905 0.910 0.870 0.889 test Discussion Postoperative pneumonia, a severe complication following lower limb fracture surgery, frequently leads to severe adverse outcomes[12]. To achieve more reliable and applicable results, this study incorporated a comprehensive assessment of risk factors from a large sample. The incidence of postoperative pneumonia was found to be 6.44%, with sex, age, smoking history, alcohol consumption, operation duration, cerebrovascular disease, hypertension, diabetes, fracture type, surgical grade, globulin ratio, platelet count, and C-reactive protein levels identified as significant predictors. In this study, age emerged as one of the independent factors influencing postoperative pulmonary infections[13, 14]. Previous research has highlighted that the age of the patient at the time of injury is the most crucial factor affecting prognosis, with older age inversely proportional to survival time and directly proportional to mortality[15, 16]. Elderly individuals exhibit decreased phagocytic capacity of macrophages in response to bacterial invasion and inflammatory stimuli, leading to reduced clearance of respiratory secretions[17]. This results in the accumulation of secretions and immune dysregulation in lung tissue, thereby increasing sensitivity to hospital-acquired pathogens. Additionally, with alveolar enlargement and changes in lung parenchyma, aging leads to gradual decline in lung function, atrophy of nasal and bronchial mucosa, facilitating bacterial colonization in the lungs and elevating the risk of pneumonia[18, 19]. A retrospective analysis by Bohl et al. of 29,377 hip fracture patients from 2006 to 2014 revealed age, preoperative COPD, exertional dyspnea, anemia, and inability to self-care as independent risk factors for postoperative pneumonia[20]. A prospective study with lower extremity fractures found that advanced age, diabetes, anemia, history of stroke, number of internal comorbidities, ASA anesthesia grade, as well as hypoproteinemia, high serum creatinine, and red blood cell distribution width were independent risk factors for postoperative pneumonia[16]. Henderson et al.'s research underscored the highest predictive value of preoperative COPD for postoperative respiratory complications, which also increased one-year mortality[21]. Elderly patients with hip fractures often have comorbidities such as hypertension, coronary heart disease, diabetes, and stroke, coupled with poor nutritional status, leading to a high risk of postoperative complications. However, these patients often exhibit no clinical symptoms in the early stages of pulmonary infection, posing significant challenges to clinical diagnosis and treatment. Compared to other published prediction models[22–24], the predictive model in this study incorporates objective indicators such as age, gender, RBC and C-reactive protein, ensuring consistent assessments by healthcare professionals with varying clinical experiences. In our research, we found that the XGBoost model stands out among many machine learning models, but the performance of the transformer model seems to be more significant, but the previous transformer model is often applied to the processing of natural language and sequence data. This attempt in clinical text data provides a direction for future research. Based on our findings, patients undergoing surgery for lower extremity fractures should undergo thorough screening of baseline conditions (smoking history, alcohol consumption, stroke, diabetes, hypertension), and corresponding intervention plans should be implemented. Clinical nurses should closely monitor laboratory test results and provide personalized nursing and therapeutic interventions for patients with abnormal results. Nevertheless, this study has limitations, including its retrospective nature and potential information bias in data collection. Furthermore, external validation of the model was not conducted. Conclusions In this study, machine learning algorithm such as SVM, KNN, MLP, ET, RF, XGBoost, LightGBM, AdaBoost and deep learning model Transformer were used to construct models to predict postoperative pneumonia in elderly people with hip fracture. Our study proposes XGBoost and Transformer algorithm models that can guide decisions regarding primary prevention of postoperative pneumonia in patients with lower limb fractures. After validation, these models show strong predictive power. Taken together, these models have important potential as a valuable tool for the prevention and management of postoperative pneumonia in patients with lower limb fractures. Declarations Funding: This study was supported by Research Project of Nantong Health Commission (MSZ2023012). Author Contributions: YQC, MXM, DDQ and CXX contributed to the study conception and design. Data collection and statistical analysis were performed by YQC, MXM, DDQ. The first draft of the manuscript was written by YQC. Funding was obtained by CXX. All authors have read and agreed to the published version of the manuscript. Clinical trial number: Not applicable. Ethics approval and consent to participate: This study includes retrospective data use (2013-2021) and prospective data collection (2021-2024), and was approved by the Institutional Review Committee of the Affiliated Hospital of Nantong University in 2021 (No. 2021-K153-01). The Institutional Review Committee of the Affiliated Hospital of Nantong University waived informed consent for anonymous historical data and mandatory written consent for prospective data (Institutional Review Board Affiliated Hospital of Nantong University No. 2021-K153-01). All the procedures followed were in accordance with the ethical guidelines of the Declaration of Helsinki. Consent for publication: Not applicable. Competing interests: The authors declare that they have no conflict of interest. Data Availability Statement: The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request. References Lee SH, Kim KU: Risk Factors for Postoperative Pneumonia in the Elderly Following Hip Fracture Surgery: A Systematic Review and Meta-Analysis . GERIATR ORTHOP SURG 2022, 13 :1771175247. de Haan E, Roukema GR, van Rijckevorsel V, Kuijper TM, de Jong L: Risk Factors for 30-Days Mortality After Proximal Femoral Fracture Surgery, a Cohort Study . CLIN INTERV AGING 2024, 19 :539-549. Mizera MM, Putur D, Tarasova A, Gjonbalaj E, Seref-Ferlengez Z, Munoz AM, Akioyamen NO, Kahn M: Increased 90-Day Morbidity and Mortality Among Patients With Hip Fracture During the COVID-19 Pandemic . ORTHOPEDICS 2024:1-6. Bohl DD, Sershon RA, Saltzman BM, Darrith B, Della VC: Incidence, Risk Factors, and Clinical Implications of Pneumonia After Surgery for Geriatric Hip Fracture . J ARTHROPLASTY 2018, 33 (5):1552-1556. Zhao K, Zhang J, Li J, Guo J, Meng H, Zhu Y, Zhang Y, Hou Z: In-Hospital Postoperative Pneumonia Following Geriatric Intertrochanteric Fracture Surgery: Incidence and Risk Factors . CLIN INTERV AGING 2020, 15 :1599-1609. Han SB, Kim SB, Shin KH: Risk factors for postoperative pneumonia in patients undergoing hip fracture surgery: a systematic review and meta-analysis . BMC MUSCULOSKEL DIS 2022, 23 (1):553. Cheng X, Liu S, Yan J, Wang L, Lei X, Wu H, Zhu Y, Zhang Y: The Time-Effect Relationship between Time to Surgery and In-Hospital Postoperative Pneumonia in Older Patients with Hip Fracture . GERONTOLOGY 2024, 70 (2):155-164. Guo J, He Q, Peng C, Dai R, Li W, Su Z, Li Y: Machine learning algorithms to predict risk of postoperative pneumonia in elderly with hip fracture . J ORTHOP SURG RES 2023, 18 (1):571. Dai A, Liu H, Shen P, Feng Y, Zhong Y, Ma M, Hu Y, Huang K, Chen C, Xia H et al : Incorporating preoperative frailty to assist in early prediction of postoperative pneumonia in elderly patients with hip fractures: an externally validated online interpretable machine learning model . BMC GERIATR 2024, 24 (1):472. Kobes T, Smeeing D, Hietbrink F, Benders K, Houwert RM, van Baal M: Definitions of hospital-acquired pneumonia in trauma research: a systematic review . EUR J TRAUMA EMERG S 2024. Zhang Y, Li Q, Xin Y: Research on eight machine learning algorithms applicability on different characteristics data sets in medical classification tasks . FRONT COMPUT NEUROSC 2024, 18 :1345575. Ahn J, Chang JS, Kim JW: Postoperative Pneumonia and Aspiration Pneumonia Following Elderly Hip Fractures . J NUTR HEALTH AGING 2022, 26 (7):732-738. Dai Y, Sun G, Hu H, Wang C, Wang H, Zha Y, Sheng Y, Hou J, Bian J, Bo L: Risk factors for postoperative pulmonary complications in elderly patients receiving elective colorectal surgery: A retrospective study . FRONT ONCOL 2022, 12 :1002025. Mimae T, Miyata Y, Kumada T, Handa Y, Tsutani Y, Okada M: Interstitial pneumonia and advanced age negatively influence postoperative pulmonary function . INTERACT CARDIOV TH 2022, 34 (5):753-759. Lee JH, Kim MJ, Hong JY, Myung J, Roh YH, Chung SP: The elderly age criterion for increased in-hospital mortality in trauma patients: a retrospective cohort study . SCAND J TRAUMA RESUS 2021, 29 (1):133. Huang J, Ge H, Zhu X, Xue C, Su Q, Chen X, Cheng B: Risk factors analysis and nomogram construction for postoperative pulmonary infection in elderly patients with hip fractures . AGING CLIN EXP RES 2023, 35 (9):1891-1899. Enne VI, Aydin A, Baldan R, Owen DR, Richardson H, Ricciardi F, Russell C, Nomamiukor-Ikeji BO, Swart AM, High J et al : Multicentre evaluation of two multiplex PCR platforms for the rapid microbiological investigation of nosocomial pneumonia in UK ICUs: the INHALE WP1 study . THORAX 2022, 77 (12):1220-1228. Cho SJ, Stout-Delgado HW: Aging and Lung Disease . ANNU REV PHYSIOL 2020, 82 :433-459. Schneider JL, Rowe JH, Garcia-de-Alba C, Kim CF, Sharpe AH, Haigis MC: The aging lung: Physiology, disease, and immunity . CELL 2021, 184 (8):1990-2019. Bohl DD, Sershon RA, Saltzman BM, Darrith B, Della VC: Incidence, Risk Factors, and Clinical Implications of Pneumonia After Surgery for Geriatric Hip Fracture . J ARTHROPLASTY 2018, 33 (5):1552-1556. Miskovic A, Lumb AB: Postoperative pulmonary complications . BRIT J ANAESTH 2017, 118 (3):317-334. Russotto V, Sabate S, Canet J: Development of a prediction model for postoperative pneumonia: A multicentre prospective observational study . EUR J ANAESTH 2019, 36 (2):93-104. Xiang G, Dong X, Xu T, Feng Y, He Z, Ke C, Xiao J, Weng YM: A Nomogram for Prediction of Postoperative Pneumonia Risk in Elderly Hip Fracture Patients . RISK MANAG HEALTHC P 2020, 13 :1603-1611. Zhang X, Shen ZL, Duan XZ, Zhou QR, Fan JF, Shen J, Ji F, Tong DK: Postoperative Pneumonia in Geriatric Patients With a Hip Fracture: Incidence, Risk Factors and a Predictive Nomogram . GERIATR ORTHOP SURG 2022, 13 :1771175248. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 01 Jul, 2025 Read the published version in Scientific Reports → Version 1 posted Reviews received at journal 15 Apr, 2025 Reviewers agreed at journal 15 Apr, 2025 Reviewers invited by journal 15 Apr, 2025 Submission checks completed at journal 27 Mar, 2025 First submitted to journal 19 Mar, 2025 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-6056672","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":443164479,"identity":"e5dbf607-a6c4-467a-a61f-9b0fcb0af6a7","order_by":0,"name":"Yiqun Chen","email":"","orcid":"","institution":"The Second Affiliated Hospital of Nantong University","correspondingAuthor":false,"prefix":"","firstName":"Yiqun","middleName":"","lastName":"Chen","suffix":""},{"id":443164480,"identity":"2723cf42-11ff-4f44-a26c-6aca81953e9c","order_by":1,"name":"Mingxuan Ma","email":"","orcid":"","institution":"Affiliated Hospital of Nantong University, Medical School of Nantong University","correspondingAuthor":false,"prefix":"","firstName":"Mingxuan","middleName":"","lastName":"Ma","suffix":""},{"id":443164481,"identity":"fb9e42fc-0f30-473b-a894-ceddaee149f0","order_by":2,"name":"Dandan Qu","email":"","orcid":"","institution":"Affiliated Hospital of Nantong University, Medical-Nursing School of Nantong University","correspondingAuthor":false,"prefix":"","firstName":"Dandan","middleName":"","lastName":"Qu","suffix":""},{"id":443164482,"identity":"15014fef-f6cd-439c-97f1-b9dc417db1d9","order_by":3,"name":"Chunxiang Xu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA7ElEQVRIiWNgGAWjYDACCRBhAGYyPv5RISEnT4oWZmOGMxbGhg1EaYEANmnGtopEhgMEdPDP7jF7zFNglycfkWNsXDhPIoGxgfnhoxv4LLlzxtxwhkFyseGZM4aPZ26TyGNnYDM2zsGjxUAix0zigwFz4sb2HmMD3m0SxYwNPGzSBLUkGNQnbmzmMZPgnSOR2HCAGC0fDA4nzmfvMZPmbSBCi8SNtDLJGQbHEzfwHCs2nHFMwtiwmYBf+Gckb5Pm+VOdOH9G8sYHH2rq5OTZmx8+xqcF4cIDMBYzMcpBQL6BWJWjYBSMglEw4gAA5DZGs/U1+c0AAAAASUVORK5CYII=","orcid":"","institution":"Affiliated Hospital of Nantong University, Medical-Nursing School of Nantong University","correspondingAuthor":true,"prefix":"","firstName":"Chunxiang","middleName":"","lastName":"Xu","suffix":""}],"badges":[],"createdAt":"2025-02-18 13:38:06","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6056672/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6056672/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-04623-y","type":"published","date":"2025-07-01T15:58:52+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":80800882,"identity":"42e0aefe-b804-41a0-aa21-8b13d51c7a21","added_by":"auto","created_at":"2025-04-17 08:34:47","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1928622,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePredictors selection in the LASSO model used 10-fold cross-validation via minimum criteria. A \u003c/strong\u003eThe trend chart with a variance fluctuation coefficient. Each color curve represents a trend with a synergistic variance.\u003cstrong\u003e B \u003c/strong\u003eCross-validation result chart.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6056672/v1/e9c3103385dd37c9f58d34a9.png"},{"id":80800880,"identity":"0d23c8b7-6b5e-4143-8fe7-db1ac3c4efd9","added_by":"auto","created_at":"2025-04-17 08:34:47","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1470774,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePerformance Evaluation of the Optimal Two Models in test set: XGBoost and Transformer models.\u003c/strong\u003e \u003cstrong\u003eA\u003c/strong\u003e ROC curves in the test set. \u003cstrong\u003eB\u003c/strong\u003e Cohort test IDI between the two models in test set. \u003cstrong\u003eC\u003c/strong\u003e DCA curves in the test set. \u003cstrong\u003eD\u003c/strong\u003e Calibration curve in the test set.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6056672/v1/f77819f1e0a66cbaa2e70a04.png"},{"id":80801908,"identity":"5739bf29-b55c-48b5-9243-2826a8d4d938","added_by":"auto","created_at":"2025-04-17 08:42:47","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":2005357,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSummary plot of SHAP values for the model constructed by XGBoost algorithm. The color of each SHAP value point indicates whether the observed value is higher (red) or lower (blue). A \u003c/strong\u003eEigendensity scatter plot: Beeswarm plot.\u003cstrong\u003eB\u003c/strong\u003e Single-sample feature impact map: waterfall plot.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-6056672/v1/fdc46fe5ce10601ef6309486.png"},{"id":86179655,"identity":"eb1245fc-ff1b-4cc0-b8a2-f41a4d37ee67","added_by":"auto","created_at":"2025-07-07 16:18:07","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":6969267,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6056672/v1/4d3549bd-00d1-4b77-a224-3ea46c905903.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Machine Learning and Transformer models for Prediction of Postoperative Pneumonia Risk in Patients with Lower Limb Fractures","fulltext":[{"header":"Introduction","content":"\u003cp\u003ePostoperative pneumonia is a prevalent complication following lower extremity fracture surgery, with an incidence ranging from 5.1\u0026ndash;14.9%[1]. The increasing number of such surgeries over the past few decades has underscored the significance of this complication within the healthcare and social support systems. Epidemiological studies indicate that postoperative comorbid pneumonia in patients with lower extremity fractures can lead to a mortality rate of 27\u0026ndash;43%, extend hospitalization by 56%[2, 3], and escalate the risk of readmission by a factor of eight. These adverse outcomes are often attributed to factors such as advanced age, comorbidities, surgical trauma, prolonged limb immobilization, and diminished immune and pulmonary function[4\u0026ndash;6]. Consequently, identifying modifiable risk factors for the development of postoperative pneumonia in this patient population is crucial for improving patient outcomes[7].\u003c/p\u003e \u003cp\u003eAccurately identifying high-risk patients necessitates the determination of significant predictors of postoperative pneumonia, a complex outcome influenced by a myriad of interacting factors. The occurrence of postoperative pneumonia exhibits considerable diversity and instability, with its risk factors continually under investigation. Prior research has extensively delved into potential risk factors for postoperative pneumonia complicating lower extremity fractures, encompassing variables such as gender, age, anemia, surgical duration, hospital stay length, and select laboratory biomarkers[6]. These studies have enhanced our comprehension of factors that elevate the risk of postoperative pneumonia. However, they are constrained by limitations like small sample sizes, incomplete consideration of risk factors, and inclusion of select, older populations, thereby impeding the generalizability of their findings and conclusions.\u003c/p\u003e \u003cp\u003eThe disease prediction model, a quantitative instrument for assessing disease risk, has emerged as a crucial auxiliary in identifying high-risk populations during clinical diagnosis, treatment, and nursing care. The essence of achieving precise disease prediction lies in the thorough examination of the interplay between disease-related factors. Nevertheless, the variables incorporated in existing postoperative lung infection risk models for lower extremity fractures, as documented in the literature, are largely based on personal experience or subjective consultations, encompassing elements such as smoking history and the number of preoperative comorbidities[8, 9]. Additionally, these models may incorporate intricate scoring tools, such as functional status assessments, which can pose challenges for clinical caregivers in terms of practical application. Consequently, we conducted a retrospective analysis of objective risk factors for postoperative pneumonia following lower extremity fractures. Leveraging machine learning and deep learning algorithms, we developed an advanced and intelligent tool to aid clinical caregivers in the early prevention of postoperative pneumonia.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy Population\u003c/h2\u003e \u003cp\u003eThis retrospective cohort study examined two distinct patient groups (pneumonia and non-pneumonia) who had undergone surgery for lower limb fractures at the Affiliated Hospital of Nantong University. The study population encompassed individuals with fractures in the hip, femur, tibia, and fibula. The medical records of these patients were retrospectively analyzed over an eight-year period, spanning from January 2015 to January 2023, yielding a total of 4,424 cases for evaluation. Randomly divided into training dataset (n\u0026thinsp;=\u0026thinsp;3539) and test set (n\u0026thinsp;=\u0026thinsp;1065) according to 8:2 data.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eInclusion and Exclusion\u003c/h3\u003e\n\u003cp\u003eInclusion criteria encompassed individuals diagnosed with lower limb fractures, encompassing tibial, femoral, and pelvic fractures, based on a comprehensive assessment of clinical symptoms, signs, and imaging findings. Exclusion criteria included those with a preexisting diagnosis of pneumonia prior to surgery, pathological fractures, or combined fractures in other anatomical regions, such as ribs, upper limb bones, or vertebrae. Additionally, patients who expired for any reason during hospitalization were excluded from the study.\u003c/p\u003e \u003cp\u003eTo assess the development of postoperative pneumonia, researchers thoroughly reviewed medical records from the day subsequent to surgery until hospital discharge. A rigorous data entry process involving double entry and cross-checking was implemented to minimize potential errors. Any discrepancies encountered during this process were resolved through collaborative discussion and consensus among the research team.\u003c/p\u003e\n\u003ch3\u003eDefinition of Pneumonia\u003c/h3\u003e\n\u003cp\u003e The diagnosis of pneumonia in this study adhered to the guidelines established by the American Thoracic Society for healthcare-associated pneumonia, augmented by relevant online resources[10]. The diagnostic criteria encompassed the presence of novel, progressive, or persistent respiratory symptoms, inclusive of coughing and the production of purulent secretions. Additionally, patients exhibited either fever (body temperature exceeding 38\u0026deg;C) or hypothermia (body temperature falling below 36\u0026deg;C). Confirmation of lung consolidation or moist rales through physical examination, along with laboratory findings indicative of leukocytosis (white blood cell count exceeding 10\u0026times;109/L) or leukopenia (white blood cell count falling below 4\u0026times;109/L), further strengthened the diagnosis. Finally, positive results from blood cultures or sputum samples provided definitive evidence of pneumonia.\u003c/p\u003e\n\u003ch3\u003eFeature Selection\u003c/h3\u003e\n\u003cp\u003eWe identified potential predictors of pneumonia formation by selecting factors that significantly impact its development. This was accomplished through a comprehensive literature review and clinical expertise. The resulting list of predictors included age, gender,Fracture type, VTE, diabetes, hypertension, COPD, cancer, heart disease, cerebrovascular disease, atrial fibrillation, hypoalbuminemia, total cholesterol, Free fatty acid, Albumin, Albumin to globulin ratio, calcium, potassium, Platelet, red blood cell, leukocyte, Fibrinogen, D-dimer, Length of stay, smoking and alcohol consumption history, operation time, blood group, Surgical grade and C-reactive protein levels. Operation time was dichotomized based on a threshold of 3 hours.\u003c/p\u003e \u003cp\u003eTo assess the correlation between features with high repeatability, Spearman's rank correlation coefficient was utilized (as depicted in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, which shows the Spearman correlation of each feature). In cases where the correlation coefficient exceeded 0.9 between any two features, only one of the highly correlated features was retained to avoid redundancy. To maximize the representation of features while minimizing redundancy, a greedy recursive deletion strategy was employed for feature filtering. Specifically, the feature exhibiting the greatest redundancy within the current set was iteratively removed. Ultimately, this process resulted in the retention of 23 features.\u003c/p\u003e \u003cp\u003eThe Least Absolute Shrinkage and Selection Operator (LASSO) regression model was utilized on the discovery dataset to construct a radiomics signature. By adjusting the regularization parameter λ, LASSO shrinks all regression coefficients towards zero, effectively eliminating the coefficients of irrelevant features by setting them exactly to zero. To identify an optimal λ, we employed 10-fold cross-validation with a minimum error criterion, ensuring that the final λ value minimized the cross-validation error. The retained features with nonzero coefficients were then incorporated into a regression model, forming the clinical signature. Subsequently, we derived a clinical score for each patient by linearly combining the retained features, with each feature weighted by its corresponding model coefficient. This analysis was facilitated by the Python scikit-learn package for LASSO regression modeling.\u003c/p\u003e\n\u003ch3\u003eSMOTE-ENN Resampling\u003c/h3\u003e\n\u003cp\u003eIn order to solve the problem of class imbalance in the data set, we implemented the SMOTE-ENN hybrid resampling technique in the training set before the model development. After resampling, the ratio: adjusted minority: majority is 1:1.2 (1:14.5 than the original). At the same time strictly ensure the independence of test set data. This method combines the synthetic minority oversampling technique (SMOTE) for minority class expansion and the edited nearest neighbor (ENN) for most class noise reduction, effectively balancing the class distribution while improving the clarity of decision boundaries.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eModel building\u003c/h2\u003e \u003cp\u003eIn the process of constructing predictive models using machine learning algorithms, clinical variables that have undergone screening are incorporated as parameter features. Eight machine learning algorithms[11], including Support Vector Machines (SVM), K-Nearest Neighbors (KNN), Multilayer Perceptron (MLP), ExtraTrees (ET), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Adaptive Boosting (AdaBoost), are employed to establish these models. To ensure the stability and reproducibility of model performance, 10-fold cross-validation resampling is utilized.\u003c/p\u003e \u003cp\u003eIn contrast, in the Transformer model, all features are incorporated for deep learning training without prior screening to select the optimal model.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eStatistical Analysis\u003c/h2\u003e \u003cp\u003eStatistical analysis was performed using SPSS 26.0 (IBM, Armonk, NY, USA). The incidence of postoperative pneumonia was determined by dividing the number of patients who developed pneumonia during hospitalization by the total number of patients included in the study. Continuous variables with normal distribution were expressed as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation, while those with non-normal distribution were expressed as medians (interquartile range), following confirmation of normality using the Shapiro-Wilk test. Subsequently, either a t-test or Whitney-U test was employed as appropriate. Categorical variables were expressed as numbers with percentages, and group differences were analyzed using Chi-square or Fisher's exact test. Statistical significance was set at \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05. To assess the predictive effectiveness of the column-line diagrams, ROC curves were constructed. The clinical utility of the predictive models was evaluated using DCA decision curves, and calibration curves were generated to assess the calibration efficiency of the column-line diagrams.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eClinical Characteristics\u003c/h2\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e summarizes the patient characteristics in the training set. We reviewed the medical records of 3311 patients without pneumonia and 228 patients with pneumonia, yielding a pneumonia incidence rate of 6.44%. Statistical significance was observed between the no pneumonia and pneumonia groups for age (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), Gender (P\u0026thinsp;=\u0026thinsp;0.803), fracture type (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), VTE (P\u0026thinsp;=\u0026thinsp;0.766), Diabetes (P\u0026thinsp;=\u0026thinsp;0.0948) Hypertension (P\u0026thinsp;=\u0026thinsp;0.96), COPD (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), Cancer (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), Heart disease (P\u0026thinsp;=\u0026thinsp;0.517), Atrial fibrillation (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), Cerebrovascular disease (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), Nephropathy (P\u0026thinsp;=\u0026thinsp;0.934), Hypoalbuminemia (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), Total cholesterol (P\u0026thinsp;=\u0026thinsp;0.971), Free fatty acid (P\u0026thinsp;=\u0026thinsp;0.0146), Albumin(P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), Albumin to globulin ratio (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), calcium (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), potassium (P\u0026thinsp;=\u0026thinsp;0.969), Platelet (P\u0026thinsp;=\u0026thinsp;0.13), Red blood cells (P\u0026thinsp;=\u0026thinsp;0.983), Leukocyte (P\u0026thinsp;=\u0026thinsp;0.971), Fibrinogen(P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), D-dimer (P\u0026thinsp;=\u0026thinsp;0.001), Length of stay (P\u0026thinsp;=\u0026thinsp;0.781), Operation time (P\u0026thinsp;=\u0026thinsp;0.382), Alcohol disease (P\u0026thinsp;=\u0026thinsp;0.190), smoking disease (P\u0026thinsp;=\u0026thinsp;0.162), CRP (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), Blood group (P\u0026thinsp;=\u0026thinsp;0.001) and Surgical grade (P\u0026thinsp;=\u0026thinsp;0.395) .\u003c/p\u003e \u003cp\u003eTo establish the model, we employed a least absolute shrinkage and selection operator (LASSO) logistic regression model, selecting only nonzero coefficients. The coefficients and mean squared error (MSE) from 10-fold cross-validation are depicted in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Nineteen non-zero coefficients were identified, including Age, Gender, Fracture type, VTE, Hypertension, COPD, Cancer, Atrial fibrillation, Cerebrovascular disease, Hypoalbuminemia, Free fatty acid, Albumin, Albumin to globulin ratio, Calcium, Fibrinogen, D-dimer, Alcohol, Surgical grade and CRP.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eComparison of machine learning model and deep learning model\u003c/h2\u003e \u003cp\u003eEight machine learning models and a deep learning model were used to evaluate the performance of the test set. As can be seen from Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, the AUC of the SVM model is 0.764, F1-score is 0.666, the AUC of the KNN model is 0.812, F1-score is 0.755, the AUC of the random forest model is 0.834, F1-score is 0.760, the AUC of the ExtraTrees model is 0.843, F1-score is 0.768, the AUC of the LightGBM model is 0.869, F1-score is 0.778, the AUC of the AdaBoost model is 0.852, F1-score is 0.768, the AUC of the MLP model is 0.859, F1-score is 0.763, and the AUC of the XGBoost model is 0.866, F1-score is 0.807, the AUC of the Transformer model is 0.946, and the F1-score is 0.889. The XGBoost model and the Transformer model have the best performance among many models.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e shows the ROC curve, decision curve and calibration curve of the two optimal models in the test set. It can also be seen from Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA that both the XGBoost and transformer models show superior performance, with AUC values of 0.866 and 0.946, respectively. From Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB, the transformer model dominates in the same cohort study. Moreover, the decision curve area of the transformer model is the largest in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC. In addition, from Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD, the oblique of the transformer model calibration curve is close to 1, indicating that the risk of postoperative pneumonia predicted by the transformer model is in good agreement with the actual risk.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eModel Visualization\u003c/h2\u003e \u003cp\u003eWe further plotted a summary plot of SHAP values to interpret the XGBoost model results (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). For each feature, a point corre sponds to a patient, and the position of the point on the x-axis indicates the effect of the feature on the model output for that particular patient. In Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA, the SHAP values of all indicators are labeled in color (red/blue) to ensure that the contribution direction is clear at a glance. Specifically, age, gender, albumin level, calcium ion level, and history of hypertension occupy a high weight in all characteristics. Among them, albumin and calcium ion levels are negatively correlated, and others are positively correlated.\u003c/p\u003e \u003cp\u003eIn the individual prediction process, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB, the current patient's age is 82 and the SHAP value is +\u0026thinsp;0.67. Advanced age is significantly positively correlated with the risk of pneumonia events; albumin level (32.56 g/L, lower than the cohort average) contributes to the SHAP value of +\u0026thinsp;0.14, indicating that hypoalbuminemia increases the risk of pneumonia; calcium ion level is 2.026 (lower than the cohort average), and the SHAP value is +\u0026thinsp;0.16, indicating that low calcium also enhances the risk of events. The corresponding SHAP values of other indicators such as history of hypertension, cerebrovascular disease, high D-dimer, and high CRP all indicate that the risk of events is increased. For gender, women tend to have a lower risk of pneumonia after fracture surgery.\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\u003eClinical characters in the training set.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\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=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c4\" namest=\"c2\"\u003e \u003cp\u003eTrain\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo pneumonia (N\u0026thinsp;=\u0026thinsp;3311)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePneumonia (N\u0026thinsp;=\u0026thinsp;228)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003eP\u003c/em\u003e-value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAge\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\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 (range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e60.0 [2.00, 102]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e75.0 [12.0, 96.0]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eGender\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.803\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1765 (53.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e129 (56.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1546 (46.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e99.0 (43.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eFracture type\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\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\u003eHip\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e147 (4.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e6.00 (2.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemur\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1996 (60.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e167 (73.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTibial\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1005 (30.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e24.0 (10.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMultiple\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e163 (4.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e31.0 (13.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eVTE\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.766\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=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3157 (95.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e215 (94.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e154 (4.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e13.0 (5.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eDiabetes\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.0948\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=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2888 (87.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e192 (84.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e423 (12.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e36.0 (15.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eHypertension\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.96\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=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2412 (72.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e162 (71.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e899 (27.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e66.0 (28.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCOPD\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\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\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3221 (97.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e200 (87.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e90.0 (2.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e28.0 (12.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCancer\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\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\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3118 (94.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e200 (87.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e193 (5.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e28.0 (12.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eHeart disease\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.517\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=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3156 (95.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e214 (93.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e155 (4.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e14.0 (6.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAtrial fibrillation\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\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\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3168 (95.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e208 (91.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e143 (4.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e20.0 (8.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCerebrovascular disease\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\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\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2943 (88.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e188 (82.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e368 (11.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e40.0 (17.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eNephropathy\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.934\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=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3248 (98.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e222 (97.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e63.0 (1.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e6.00 (2.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eHypoalbuminemia\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\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\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3177 (96.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e209 (91.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e134 (4.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e19.0 (8.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eTotal cholesterol\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.971\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian (range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3.74 [0.700, 7.91]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.74 [1.92, 7.14]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eFree fatty acid\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.0146\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian (range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.510 [0.0100, 4.09]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.550 [0.0400, 1.94]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAlbumin\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\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 (range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e37.0 [13.3, 54.3]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e34.3 [18.9, 48.6]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAlbumin to globulin ratio\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\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 (range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.42 [0.370, 3.25]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.33 [0.670, 2.20]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003ecalcium\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\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 (range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2.24 [1.70, 3.04]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.17 [0.930, 2.63]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003epotassium\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.969\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian (range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3.88 [2.00, 7.43]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.91 [2.41, 8.25]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003ePlatelet\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.13\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian (range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e186 [13.0, 1470]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e167 [23.0, 1220]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eRed blood cell\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.983\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian (range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3.83 [0, 2880]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.62 [1.25, 392]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eLeukocyte\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.971\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian (range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e8.04 [1.34, 1000]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e8.69 [2.54, 38.4]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eFibrinogen\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\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 (range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3.33 [0.300, 11.5]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.64 [0.950, 7.83]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eD dimer\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.00106\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian (range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5.88 [0.100, 103]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e7.81 [0.650, 153]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eLength of stay\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.781\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian (range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e12.0 [1.00, 93.0]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e13.0 [1.00, 59.0]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eOperation time\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.382\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026le;\u0026thinsp;3h\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3030 (91.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e206 (90.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026gt;3h\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e281 (8.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e22.0 (9.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAlcohol\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.19\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=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3225 (97.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e217 (95.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e86.0 (2.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e11.0 (4.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eSmoking\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.162\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=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3211 (97.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e216 (94.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e100 (3.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e12.0 (5.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCRP\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\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 (range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e41.4 [0.250, 333]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e51.6 [0.500, 200]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eBlood group\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.00162\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1059 (32.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e63.0 (27.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eB\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1296 (39.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e101 (44.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAB\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e724 (21.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e45.0 (19.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e232 (7.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e19.0 (8.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eSurgical grade\u003c/b\u003e\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.0395\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e21.0 (0.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0 (0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eII\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e119 (3.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.00 (1.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIII\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2513 (75.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e169 (74.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIV\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e658 (19.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e56.0 (24.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \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\u003ePerformance Demonstration of Machine Learning Models and Transformer Model.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"10\"\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 \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eACC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAUC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e95%CI\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003ePrecision\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eRecall\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eF1\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003eTask\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.752\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.820\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8097\u0026ndash;0.8310\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.708\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.788\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.731\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.708\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.719\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etrain\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.705\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.767\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.7428\u0026ndash;0.7904\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.655\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.745\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.676\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.655\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.666\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etest\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eKNN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.774\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.829\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8015\u0026ndash;0.8557\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.684\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.872\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.854\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.684\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.760\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etrain\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eKNN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.760\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.812\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8004\u0026ndash;0.8241\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.713\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.810\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.802\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.713\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.755\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etest\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRandomForest\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.769\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.837\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8107\u0026ndash;0.8626\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.734\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.809\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.807\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.734\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.769\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etrain\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRandomForest\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.759\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.834\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8231\u0026ndash;0.8445\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.733\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.787\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.788\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.733\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.760\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etest\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExtraTrees\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.794\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.856\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8313\u0026ndash;0.8799\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.846\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.738\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.779\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.846\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.811\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etrain\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExtraTrees\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.764\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.843\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8321\u0026ndash;0.8529\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.753\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.776\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.783\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.753\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.768\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etest\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMLP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.797\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.875\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8572\u0026ndash;0.8919\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.825\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.774\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.748\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.825\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.784\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etrain\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMLP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.783\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.859\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8496\u0026ndash;0.8683\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.781\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.784\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.746\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.781\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.763\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etest\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLightGBM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.797\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.873\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8561\u0026ndash;0.8909\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.831\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.771\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.746\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.831\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.786\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etrain\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLightGBM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.798\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.869\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8598\u0026ndash;0.8776\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.684\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.789\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.768\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.789\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.778\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etest\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAdaBoost\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.785\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.864\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8407\u0026ndash;0.8872\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.762\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.811\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.815\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.762\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.787\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etrain\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAdaBoost\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.771\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.852\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8422\u0026ndash;0.8622\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.733\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.811\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.807\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.733\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.768\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etest\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eXGBoost\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.800\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.876\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8668\u0026ndash;0.8851\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.846\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.744\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.781\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.846\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.812\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etrain\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eXGBoost\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.786\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.866\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8432\u0026ndash;0.8894\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.855\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.712\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.764\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.855\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.807\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etest\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTransformer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.908\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.964\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.9596\u0026ndash;0.9687\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.919\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.895\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.904\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.919\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.912\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etrain\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTransformer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.887\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.946\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.9308\u0026ndash;0.9609\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.870\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.905\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.910\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.870\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.889\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003etest\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\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003ePostoperative pneumonia, a severe complication following lower limb fracture surgery, frequently leads to severe adverse outcomes[12]. To achieve more reliable and applicable results, this study incorporated a comprehensive assessment of risk factors from a large sample. The incidence of postoperative pneumonia was found to be 6.44%, with sex, age, smoking history, alcohol consumption, operation duration, cerebrovascular disease, hypertension, diabetes, fracture type, surgical grade, globulin ratio, platelet count, and C-reactive protein levels identified as significant predictors. In this study, age emerged as one of the independent factors influencing postoperative pulmonary infections[13, 14]. Previous research has highlighted that the age of the patient at the time of injury is the most crucial factor affecting prognosis, with older age inversely proportional to survival time and directly proportional to mortality[15, 16]. Elderly individuals exhibit decreased phagocytic capacity of macrophages in response to bacterial invasion and inflammatory stimuli, leading to reduced clearance of respiratory secretions[17]. This results in the accumulation of secretions and immune dysregulation in lung tissue, thereby increasing sensitivity to hospital-acquired pathogens. Additionally, with alveolar enlargement and changes in lung parenchyma, aging leads to gradual decline in lung function, atrophy of nasal and bronchial mucosa, facilitating bacterial colonization in the lungs and elevating the risk of pneumonia[18, 19].\u003c/p\u003e \u003cp\u003eA retrospective analysis by Bohl et al. of 29,377 hip fracture patients from 2006 to 2014 revealed age, preoperative COPD, exertional dyspnea, anemia, and inability to self-care as independent risk factors for postoperative pneumonia[20]. A prospective study with lower extremity fractures found that advanced age, diabetes, anemia, history of stroke, number of internal comorbidities, ASA anesthesia grade, as well as hypoproteinemia, high serum creatinine, and red blood cell distribution width were independent risk factors for postoperative pneumonia[16]. Henderson et al.'s research underscored the highest predictive value of preoperative COPD for postoperative respiratory complications, which also increased one-year mortality[21].\u003c/p\u003e \u003cp\u003eElderly patients with hip fractures often have comorbidities such as hypertension, coronary heart disease, diabetes, and stroke, coupled with poor nutritional status, leading to a high risk of postoperative complications. However, these patients often exhibit no clinical symptoms in the early stages of pulmonary infection, posing significant challenges to clinical diagnosis and treatment. Compared to other published prediction models[22\u0026ndash;24], the predictive model in this study incorporates objective indicators such as age, gender, RBC and C-reactive protein, ensuring consistent assessments by healthcare professionals with varying clinical experiences.\u003c/p\u003e \u003cp\u003eIn our research, we found that the XGBoost model stands out among many machine learning models, but the performance of the transformer model seems to be more significant, but the previous transformer model is often applied to the processing of natural language and sequence data. This attempt in clinical text data provides a direction for future research.\u003c/p\u003e \u003cp\u003eBased on our findings, patients undergoing surgery for lower extremity fractures should undergo thorough screening of baseline conditions (smoking history, alcohol consumption, stroke, diabetes, hypertension), and corresponding intervention plans should be implemented. Clinical nurses should closely monitor laboratory test results and provide personalized nursing and therapeutic interventions for patients with abnormal results. Nevertheless, this study has limitations, including its retrospective nature and potential information bias in data collection. Furthermore, external validation of the model was not conducted.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eIn this study, machine learning algorithm such as SVM, KNN, MLP, ET, RF, XGBoost, LightGBM, AdaBoost and deep learning model Transformer were used to construct models to predict postoperative pneumonia in elderly people with hip fracture. Our study proposes XGBoost and Transformer algorithm models that can guide decisions regarding primary prevention of postoperative pneumonia in patients with lower limb fractures. After validation, these models show strong predictive power. Taken together, these models have important potential as a valuable tool for the prevention and management of postoperative pneumonia in patients with lower limb fractures.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding:\u0026nbsp;\u003c/strong\u003eThis study was supported by Research Project of Nantong Health Commission (MSZ2023012).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions:\u003c/strong\u003e YQC, MXM, DDQ and CXX contributed to the study conception and design. Data collection and statistical analysis were performed by YQC, MXM, DDQ. The first draft of the manuscript was written by YQC. Funding was obtained by CXX. All authors have read and agreed to the published version of the manuscript.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical trial number:\u0026nbsp;\u003c/strong\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate:\u003c/strong\u003e This study includes retrospective data use (2013-2021) and prospective data collection (2021-2024), and was approved by the Institutional Review Committee of the Affiliated Hospital of Nantong University in 2021 (No. 2021-K153-01). The Institutional Review Committee of the Affiliated Hospital of Nantong University waived informed consent for anonymous historical data and mandatory written consent for prospective data (Institutional Review Board Affiliated Hospital of Nantong University No. 2021-K153-01). All the procedures followed were in accordance with the ethical guidelines of the Declaration of Helsinki.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication:\u003c/strong\u003e Not applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests:\u0026nbsp;\u003c/strong\u003eThe authors declare that they have no conflict of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability Statement:\u003c/strong\u003e The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e"},{"header":" References","content":"\u003col\u003e\n \u003cli\u003eLee SH, Kim KU: \u003cstrong\u003eRisk Factors for Postoperative Pneumonia in the Elderly Following Hip Fracture Surgery: A Systematic Review and Meta-Analysis\u003c/strong\u003e. \u003cem\u003eGERIATR ORTHOP SURG\u003c/em\u003e 2022, \u003cstrong\u003e13\u003c/strong\u003e:1771175247.\u003c/li\u003e\n \u003cli\u003ede Haan E, Roukema GR, van Rijckevorsel V, Kuijper TM, de Jong L: \u003cstrong\u003eRisk Factors for 30-Days Mortality After Proximal Femoral Fracture Surgery, a Cohort Study\u003c/strong\u003e. \u003cem\u003eCLIN INTERV AGING\u003c/em\u003e 2024, \u003cstrong\u003e19\u003c/strong\u003e:539-549.\u003c/li\u003e\n \u003cli\u003eMizera MM, Putur D, Tarasova A, Gjonbalaj E, Seref-Ferlengez Z, Munoz AM, Akioyamen NO, Kahn M: \u003cstrong\u003eIncreased 90-Day Morbidity and Mortality Among Patients With Hip Fracture During the COVID-19 Pandemic\u003c/strong\u003e. \u003cem\u003eORTHOPEDICS\u003c/em\u003e 2024:1-6.\u003c/li\u003e\n \u003cli\u003eBohl DD, Sershon RA, Saltzman BM, Darrith B, Della VC: \u003cstrong\u003eIncidence, Risk Factors, and Clinical Implications of Pneumonia After Surgery for Geriatric Hip Fracture\u003c/strong\u003e. \u003cem\u003eJ ARTHROPLASTY\u003c/em\u003e 2018, \u003cstrong\u003e33\u003c/strong\u003e(5):1552-1556.\u003c/li\u003e\n \u003cli\u003eZhao K, Zhang J, Li J, Guo J, Meng H, Zhu Y, Zhang Y, Hou Z: \u003cstrong\u003eIn-Hospital Postoperative Pneumonia Following Geriatric Intertrochanteric Fracture Surgery: Incidence and Risk Factors\u003c/strong\u003e. \u003cem\u003eCLIN INTERV AGING\u003c/em\u003e 2020, \u003cstrong\u003e15\u003c/strong\u003e:1599-1609.\u003c/li\u003e\n \u003cli\u003eHan SB, Kim SB, Shin KH: \u003cstrong\u003eRisk factors for postoperative pneumonia in patients undergoing hip fracture surgery: a systematic review and meta-analysis\u003c/strong\u003e. \u003cem\u003eBMC MUSCULOSKEL DIS\u003c/em\u003e 2022, \u003cstrong\u003e23\u003c/strong\u003e(1):553.\u003c/li\u003e\n \u003cli\u003eCheng X, Liu S, Yan J, Wang L, Lei X, Wu H, Zhu Y, Zhang Y: \u003cstrong\u003eThe Time-Effect Relationship between Time to Surgery and In-Hospital Postoperative Pneumonia in Older Patients with Hip Fracture\u003c/strong\u003e. \u003cem\u003eGERONTOLOGY\u003c/em\u003e 2024, \u003cstrong\u003e70\u003c/strong\u003e(2):155-164.\u003c/li\u003e\n \u003cli\u003eGuo J, He Q, Peng C, Dai R, Li W, Su Z, Li Y: \u003cstrong\u003eMachine learning algorithms to predict risk of postoperative pneumonia in elderly with hip fracture\u003c/strong\u003e. \u003cem\u003eJ ORTHOP SURG RES\u003c/em\u003e 2023, \u003cstrong\u003e18\u003c/strong\u003e(1):571.\u003c/li\u003e\n \u003cli\u003eDai A, Liu H, Shen P, Feng Y, Zhong Y, Ma M, Hu Y, Huang K, Chen C, Xia H\u003cem\u003e\u0026nbsp;et al\u003c/em\u003e: \u003cstrong\u003eIncorporating preoperative frailty to assist in early prediction of postoperative pneumonia in elderly patients with hip fractures: an externally validated online interpretable machine learning model\u003c/strong\u003e. \u003cem\u003eBMC GERIATR\u003c/em\u003e 2024, \u003cstrong\u003e24\u003c/strong\u003e(1):472.\u003c/li\u003e\n \u003cli\u003eKobes T, Smeeing D, Hietbrink F, Benders K, Houwert RM, van Baal M: \u003cstrong\u003eDefinitions of hospital-acquired pneumonia in trauma research: a systematic review\u003c/strong\u003e. \u003cem\u003eEUR J TRAUMA EMERG S\u003c/em\u003e 2024.\u003c/li\u003e\n \u003cli\u003eZhang Y, Li Q, Xin Y: \u003cstrong\u003eResearch on eight machine learning algorithms applicability on different characteristics data sets in medical classification tasks\u003c/strong\u003e. \u003cem\u003eFRONT COMPUT NEUROSC\u003c/em\u003e 2024, \u003cstrong\u003e18\u003c/strong\u003e:1345575.\u003c/li\u003e\n \u003cli\u003eAhn J, Chang JS, Kim JW: \u003cstrong\u003ePostoperative Pneumonia and Aspiration Pneumonia Following Elderly Hip Fractures\u003c/strong\u003e. \u003cem\u003eJ NUTR HEALTH AGING\u003c/em\u003e 2022, \u003cstrong\u003e26\u003c/strong\u003e(7):732-738.\u003c/li\u003e\n \u003cli\u003eDai Y, Sun G, Hu H, Wang C, Wang H, Zha Y, Sheng Y, Hou J, Bian J, Bo L: \u003cstrong\u003eRisk factors for postoperative pulmonary complications in elderly patients receiving elective colorectal surgery: A retrospective study\u003c/strong\u003e. \u003cem\u003eFRONT ONCOL\u003c/em\u003e 2022, \u003cstrong\u003e12\u003c/strong\u003e:1002025.\u003c/li\u003e\n \u003cli\u003eMimae T, Miyata Y, Kumada T, Handa Y, Tsutani Y, Okada M: \u003cstrong\u003eInterstitial pneumonia and advanced age negatively influence postoperative pulmonary function\u003c/strong\u003e. \u003cem\u003eINTERACT CARDIOV TH\u003c/em\u003e 2022, \u003cstrong\u003e34\u003c/strong\u003e(5):753-759.\u003c/li\u003e\n \u003cli\u003eLee JH, Kim MJ, Hong JY, Myung J, Roh YH, Chung SP: \u003cstrong\u003eThe elderly age criterion for increased in-hospital mortality in trauma patients: a retrospective cohort study\u003c/strong\u003e. \u003cem\u003eSCAND J TRAUMA RESUS\u003c/em\u003e 2021, \u003cstrong\u003e29\u003c/strong\u003e(1):133.\u003c/li\u003e\n \u003cli\u003eHuang J, Ge H, Zhu X, Xue C, Su Q, Chen X, Cheng B: \u003cstrong\u003eRisk factors analysis and nomogram construction for postoperative pulmonary infection in elderly patients with hip fractures\u003c/strong\u003e. \u003cem\u003eAGING CLIN EXP RES\u003c/em\u003e 2023, \u003cstrong\u003e35\u003c/strong\u003e(9):1891-1899.\u003c/li\u003e\n \u003cli\u003eEnne VI, Aydin A, Baldan R, Owen DR, Richardson H, Ricciardi F, Russell C, Nomamiukor-Ikeji BO, Swart AM, High J\u003cem\u003e\u0026nbsp;et al\u003c/em\u003e: \u003cstrong\u003eMulticentre evaluation of two multiplex PCR platforms for the rapid microbiological investigation of nosocomial pneumonia in UK ICUs: the INHALE WP1 study\u003c/strong\u003e. \u003cem\u003eTHORAX\u003c/em\u003e 2022, \u003cstrong\u003e77\u003c/strong\u003e(12):1220-1228.\u003c/li\u003e\n \u003cli\u003eCho SJ, Stout-Delgado HW: \u003cstrong\u003eAging and Lung Disease\u003c/strong\u003e. \u003cem\u003eANNU REV PHYSIOL\u003c/em\u003e 2020, \u003cstrong\u003e82\u003c/strong\u003e:433-459.\u003c/li\u003e\n \u003cli\u003eSchneider JL, Rowe JH, Garcia-de-Alba C, Kim CF, Sharpe AH, Haigis MC: \u003cstrong\u003eThe aging lung: Physiology, disease, and immunity\u003c/strong\u003e. \u003cem\u003eCELL\u003c/em\u003e 2021, \u003cstrong\u003e184\u003c/strong\u003e(8):1990-2019.\u003c/li\u003e\n \u003cli\u003eBohl DD, Sershon RA, Saltzman BM, Darrith B, Della VC: \u003cstrong\u003eIncidence, Risk Factors, and Clinical Implications of Pneumonia After Surgery for Geriatric Hip Fracture\u003c/strong\u003e. \u003cem\u003eJ ARTHROPLASTY\u003c/em\u003e 2018, \u003cstrong\u003e33\u003c/strong\u003e(5):1552-1556.\u003c/li\u003e\n \u003cli\u003eMiskovic A, Lumb AB: \u003cstrong\u003ePostoperative pulmonary complications\u003c/strong\u003e. \u003cem\u003eBRIT J ANAESTH\u003c/em\u003e 2017, \u003cstrong\u003e118\u003c/strong\u003e(3):317-334.\u003c/li\u003e\n \u003cli\u003eRussotto V, Sabate S, Canet J: \u003cstrong\u003eDevelopment of a prediction model for postoperative pneumonia: A multicentre prospective observational study\u003c/strong\u003e. \u003cem\u003eEUR J ANAESTH\u003c/em\u003e 2019, \u003cstrong\u003e36\u003c/strong\u003e(2):93-104.\u003c/li\u003e\n \u003cli\u003eXiang G, Dong X, Xu T, Feng Y, He Z, Ke C, Xiao J, Weng YM: \u003cstrong\u003eA Nomogram for Prediction of Postoperative Pneumonia Risk in Elderly Hip Fracture Patients\u003c/strong\u003e. \u003cem\u003eRISK MANAG HEALTHC P\u003c/em\u003e 2020, \u003cstrong\u003e13\u003c/strong\u003e:1603-1611.\u003c/li\u003e\n \u003cli\u003eZhang X, Shen ZL, Duan XZ, Zhou QR, Fan JF, Shen J, Ji F, Tong DK: \u003cstrong\u003ePostoperative Pneumonia in Geriatric Patients With a Hip Fracture: Incidence, Risk Factors and a Predictive Nomogram\u003c/strong\u003e. \u003cem\u003eGERIATR ORTHOP SURG\u003c/em\u003e 2022, \u003cstrong\u003e13\u003c/strong\u003e:1771175248.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Postoperative pneumonia, risk factors, lower extremity fracture, machine learning model, Transformer","lastPublishedDoi":"10.21203/rs.3.rs-6056672/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6056672/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground: \u003c/strong\u003ePostoperative pneumonia, a prevalent complication arising from lower limb fracture surgery, can significantly prolong hospitalization periods and elevate mortality rates. Consequently, early prevention and identification of this condition are crucial in improving patient prognosis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eIn this study, clinical indicators pertaining to postoperative pneumonia in patients with lower limb fractures at Nantong University Hospital, spanning the years 2016 to 2023, were subjected to a analysis. The patients who encountered postoperative pneumonia subsequent to their lower limb fracture surgeries during hospitalization were categorized as the case group, whereas those who did not develop such a complication served as the control group. To forecast the likelihood of postoperative pneumonia occurrence, both machine learning and deep learning algorithms were employed.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eThe study identified Age, Gender, Fracture type, Venous thromboembolism (VTE), Hypertension, Chronic obstructive pulmonary disease (COPD), Cancer, Atrial fibrillation, Cerebrovascular disease, Hypoalbuminemia, Free fatty acid, Albumin, Albumin to globulin ratio, Calcium, Fibrinogen, D-dimer, Alcohol, Surgical grade and C-reactive protein as significant predictors of postoperative pneumonia. XGBoost and Transformer models have better performance (AUC 0.866 VS 0.966 , F1 0.807 VS 0.889), and both models have better substantial prediction ability for the occurrence of postoperative pneumonia.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion: \u003c/strong\u003eIn conclusion, XGBoost and Transformer models serve as potential tools for the prevention and treatment of postoperative pneumonia in patients with lower-extremity fractures. By adopting appropriate health management practices, the risk of developing postoperative pneumonia in this patient population may be reduced.\u003c/p\u003e","manuscriptTitle":"Machine Learning and Transformer models for Prediction of Postoperative Pneumonia Risk in Patients with Lower Limb Fractures","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-04-17 08:34:41","doi":"10.21203/rs.3.rs-6056672/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2025-04-15T09:08:36+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"198084718078264943309200202219110702624","date":"2025-04-15T09:06:40+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-04-15T08:42:43+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-03-27T07:10:37+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-03-19T16:48:38+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"c8cb73f6-3081-453f-a201-5a4be620074b","owner":[],"postedDate":"April 17th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":47181024,"name":"Health sciences/Medical research/Translational research"},{"id":47181025,"name":"Biological sciences/Computational biology and bioinformatics/Machine learning"}],"tags":[],"updatedAt":"2025-07-07T16:09:16+00:00","versionOfRecord":{"articleIdentity":"rs-6056672","link":"https://doi.org/10.1038/s41598-025-04623-y","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-07-01 15:58:52","publishedOnDateReadable":"July 1st, 2025"},"versionCreatedAt":"2025-04-17 08:34:41","video":"","vorDoi":"10.1038/s41598-025-04623-y","vorDoiUrl":"https://doi.org/10.1038/s41598-025-04623-y","workflowStages":[]},"version":"v1","identity":"rs-6056672","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6056672","identity":"rs-6056672","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}