A machine learning model for the early prediction of Gram-negative bloodstream infection in ICU patients

A machine learning model for the early prediction of Gram-negative bloodstream infection in ICU patients | 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 Research Article A machine learning model for the early prediction of Gram-negative bloodstream infection in ICU patients Ya-Ling Zhou, Hong-Ting Da, Ting-Ting Wang, Zhong-Xin Wang, Zhong-Le Cheng This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8240564/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 31 Jan, 2026 Read the published version in BMC Medical Informatics and Decision Making → Version 1 posted 11 You are reading this latest preprint version Abstract Backgroud Gram-negative bloodstream infection (GN-BSI) can induce fatal septic shock, and the increasingly severe problem of antimicrobial resistance results in high clinical mortality particularly in intensive care unit (ICU) patients. The early identification of pathogens and timely antibiotic therapy are critical for patient outcomes. However, conventional diagnostic methods like blood culture are time-consuming and can delay treatment. Furthermore, the the implementation of molecular detection techniques in routine laboratories is often hindered by high costs and technical complexity.Machine learning (ML) offers a promising alternative for early prediction of GN-BSI. This study aims to develop an early prediction model for GN-BSI by integrating clinical and laboratory parameters from ICU patients using machine learning algorithms, thereby assisting in the early diagnosis and treatment of GN-BSI. Methods This retrospective study utilized data from ICU patients admitted to the West District of the First Affiliated Hospital of Anhui Medical University between January and July 2025. Following data preprocessing and multiple imputation of missing values, the dataset was randomly divided into training and validation sets in a 7:3 ratio. Feature selection was performed using Lasso regression and multivariate logistic regression. Seven ML models were developed and evaluated based on metrics including the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, specificity,F1-score, positive predictive value (PPV), and negative predictive value (NPV). Model interpretability was further assessed using Shapley Additive Explanation (SHAP) analysis. Results This study ultimately included 405 ICU patients. Following further feature selection, four variables were identified, including deep vein catheterization, continuous renal replacement therapy (CRRT), procalcitonin, and c-reactive protein (CRP). Early prediction models for GN-BSI in ICU patients were constructed using seven machine learning algorithms. Among them, the XGBoost model demonstrated the best performance, with the AUC value of 0.898, accuracy of 88.43%, F1 score of 0.783,PPV of 85.00%, and NPV of 89.10%. SHAP bar and beeswarm plots illustrate the contribution of the four variables to the outcome. The SHAP dependency plot and force analysis provided model interpretation at the factor level and individual level, respectively. Conclusions We have successfully developed, evaluated, and interpreted a machine learning model for predicting GN-BSI in ICU patients, facilitating timely interventions and treatments. The XGBoost model holds potential for clinical reference following validation set and further refinement. Gram-negative bloodstream infection Intensive care unit Machine learning Prediction model Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Introduction Bloodstream infection (BSI) is a life-threatening systemic condition and a major global health challenge associated with high incidence and mortality[ 1 ]. In developed countries, the incidence of BSI is estimated at 100 to 200 cases per 100,000 person-years and continues to rise[ 2 ]. Gram-negative bloodstream infection (GN-BSI) has attracted considerable clinical and public health concern due to its high incidence, potential to progress to severe sepsis, considerable mortality, and the escalating challenge of antimicrobial resistance[ 3 – 5 ]. In healthcare-associated infections, particularly within the intensive care unit (ICU), Gram-negative bloodstream infections consistently account for 25%–30% of cases, posing a major challenge for both infection control and clinical management[ 6 ]. The crude mortality of ICU patients suffering from BSI is above 30%[ 2 , 7 ]. Early empirical antibiotic therapy is critical for patient survival. However, it may contribute to the development of antimicrobial resistance. Therefore, the rapid identification of pathogens is essential, as it enables clinicians to make timely and targeted adjustments to the antibiotic regimen. Currently, blood culture remains the gold standard for diagnosing bloodstream infections. Nevertheless, its time-consuming nature and high false-negative rate can delay early diagnosis and treatment[ 8 ]. Although rapid alternatives such as molecular detection techniques can guide timely antimicrobial therapy, their high cost and technical demands make them impractical for routine adoption[ 9 ]. Machine learning (ML) has demonstrated considerable potential in supporting disease diagnosis[ 10 , 11 ]. In recent years, advances in ML have led to its rapid adoption across various medical disciplines. A number of ML-based models have been successfully developed, with studies confirming their feasibility and interpretability for predicting bacteremia[ 12 , 13 ]. However, most existing machine learning models for bacteremia prediction rely on a broad set of conventional laboratory or clinical parameters.and some models incorporated a large number of features complicate the model and hinder clinical adoption. On this basis, we integrated the relatively complete clinical and laboratory parameters of patients to further improve the accuracy of model prediction[ 12 – 14 ]. The purpose of this study is to find the optimal combination of these features finally, a prediction model that can early predict GN-BSI in ICU patients is developed by using machine learning algorithm, in order to provide reference for early clinical diagnosis and treatment. Methods Study population This study employed a secondary analysis of a retrospective cohort conducted between January and July 2025 in the ICU of the West District, The First Affiliated Hospital of Anhui Medical University. The inclusion criteria were: (1) age ≥ 18 years; (2) admission to the ICU; (3) an anticipated ICU length of stay > 48 hours; and (4) at least one blood culture obtained during hospitalization. The exclusion criteria were: (1) a blood culture positive for non-gram-negative bacteria; (2) a concurrent non-Gram-negative bacterial infection identified within 7 days before or after the index Gram-negative bacteremia episode; and (3) pregnancy or lactation. Clinical and laboratory data pertaining to Gram-negative bacteremia were collected for all enrolled adult patients. To ensure statistical independence of observations, only the first episode was analyzed for patients with multiple positive blood cultures. For those with repeatedly negative cultures, a single time point was randomly selected for inclusion. A flowchart outlining the patient selection process is provided in Fig. 1. Outcome The outcome assessed was GN-BSI, defined as the growth of a gram negative bacteremia in at least one blood culture bottle[ 9 ]. Dataset We constructed datasets from the target medical centers comprising patient demographics, clinical and laboratory parameters. These variables were collected within a 24-hour window surrounding the time of blood culture collection. The dataset was structured to include the following variables: (1) patient demographics: age and sex; (2) past medical history: hypertension, diabetes mellitus, coronary heart disease, cerebrovascular disease, and solid cancer; (3) vital signs: temperature, heart rate; respiratory rate,and blood pressure; (4) antibiotic usage prior to pathogen detection; (5) traumatic operation: gastric tube insertion, urethral catheterization, endotracheal tube, tracheotomy, deep vein catheterization‌, arterial catheterization, drainage tube, continuous renal replacement therapy, and recent surgical procedures‌‌; (6) laboratory parameters:blood cells,hemagglutination,liver function,renal function; electrolytes,inflammatory markers, serum glucose, blood gas analysis ,and blood culture. Data preprocessing Data cleaning and preprocessing are essential phases in the data analysis pipeline, designed to convert raw data into a structured dataset suitable for robust statistical analysis or machine learning modeling[ 15 , 16 ]. To address missing data, we excluded variables where the proportion of missing values exceeded 15%[ 17 ]. For the remaining missing data, we employed a multiple imputation approach using the "MICE" package in R to generate complete datasets for robust analysis [ 18 ]. Feature selection A pre-seeded random number generator (123) in R software was utilized to randomly divide the cohort into training and validation sets based on a ratio of 7:3.The training sets were used for modelling,while the validation sets for internal validation. We employed an L1-penalty least absolute shrinkage and selection operator (LASSO) regression approach to screen variables, augmented with 10-fold cross-validation[[ 19 ]. LASSO regression is a method used to reduce the dimensionality of data by selecting features based on a penalty function. It effectively eliminate multicollinearity and avoid over-fitting of variables. Subsequently, we incorporate the selected variables in LASSO into a multivariate logistic regression analysis to identify the predictive variables for GN-BSI in ICU patients and construct predictive models. Machine learning algorithms Following variable selection, we employed a suite of machine learning algorithms, including Logistic Regression, Decision Tree, Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), Support Vector Machine (SVM), and Artificial Neural Network (ANN) to predict the risk of GN-BSI in ICU patients. Throughout the model development phase, we use a grid search technique and 5-fold cross validation to to tune the hyperparameters and derive the optimal model for each algorithm. The predictive performance of each model was evaluated using a comprehensive set of metrics, including accuracy, sensitivity, specificity, F1 score, positive predictive value (PPV), negative predictive value (NPV), and the area under the receiver operating characteristic curve (AUC). The optimal model was further evaluated on the validation set using a calibration curve to assess the agreement between its predictions and the actual GN-BSI outcomes, as well as by decision curve analysis (DCA) to determine its net clinical benefit. SHAP interpretability analysis The "black-box" nature of many machine learning models often obscures the influence of individual risk variables on their predictions. To enhance interpretability, we employed Shapley Additive Explanation (SHAP) values to precisely quantify the contribution and significance of each feature to the predictions generated by our optimal model[ 20 , 21 ].The higher SHAP value indicates great impact of a feature on model output. We employed a SHAP bar and beeswarm plots to evaluate feature importance, followed by utilizing SHAP dependency plot to investigate the impact of features on outcome prediction. Finally, a SHAP force analysis was used to elucidate the contribution of features in individual patients. Statistical analysis Continuous variables are presented as mean ± SEM (range) or median (interquartile range). Comparisons were conducted using either the Student’s t-test or the Wilcoxon rank-sum test. Categorical variables are expressed in terms of frequencies and percentages, with comparisons performed using the chi-square test or Fisher’s exact test, as appropriate. A two-sided P-value of less than 0.05 was deemed statistically significant. Statistical analyses were carried out utilizing SPSS version 27.0 (IBM Corp), R version 4.5.1 (The R Foundation for Statistical Computing), and Python version 3.10.4 (Python Software Foundation). Results Patient characteristics During the study period, 596 patients were admitted to ICU. After applying the exclusion criteria,which comprised patients under 18 years of age (n = 4), those with an ICU stay of less than 48 hours (n = 79), and those with blood cultures indicating non-gram-negative bacteria (n = 108),a total of 405 patients were included in the study. From this final cohort, 94 patients were identified with GN-BSI, accounting for 23.21%. Variables with a missing data ratio exceeding 15% (namely, PH, PO2, PCO2,BE, and Lac) were excluded from further analysis. The distribution of missing data for all variables is detailed in Supplementary Fig. 1 and Supplementary Table 1. The included patients were divided into the training set (284 patients) and the validation set (121 patients).In the training and validation set, the median age was 61 (IQR:49, 72) and 61.5 (IQR:53, 72) years, and 195 (62.7%) and 60 (63.83%) patients were men,respectively. A total of 81 variables were collected for each patient. The detailed features between individuals with and without GN-BSI were summarized in Table 1 . Table 1 Baseline characteristics of ICU patients with and without GN-BSI Variables Overall(n = 405) Non-GN-BSI (n = 311) GN-BSI (n = 94) p- value Male (%) 255 (62.96) 195 (62.70) 60 (63.83) 0.939 Hypertension (%) 153 (37.78) 115 (36.98) 38 (40.43) 0.629 Diabetes mellitus (%) 63 (15.56) 45 (14.47) 18 (19.15) 0.35 Coronary heart disease (%) 39 ( 9.63) 33 (10.61) 6 ( 6.38) 0.309 Cerebrovascular disease (%) 40 ( 9.88) 31 ( 9.97) 9 ( 9.57) 1 Solid cancer (%) 38 ( 9.38) 26 ( 8.36) 12 (12.77) 0.279 Recent surgical operation (%) 242 (59.75) 186 (59.81) 56 (59.57) 1 lactamase inhibitor(%) 229 (56.54) 179 (57.56) 50 (53.19) 0.529 Carbapenem antibiotics (%) 122 (30.12) 64 (20.58) 58 (61.70) < 0.001* Colistin (%) 11 ( 2.72) 2 ( 0.64) 9 ( 9.57) < 0.001* Tetracycline (%) 10 ( 2.47) 3 ( 0.96) 7 ( 7.45) 0.002* Glycopeptides antibiotics(%) 88 (21.73) 47 (15.11) 41 (43.62) < 0.001* Antifungal drugs (%) 42 (10.37) 16 ( 5.14) 26 (27.66) < 0.001* Immunosuppressive drugs (%) 18 ( 4.44) 7 ( 2.25) 11 (11.70) < 0.001* Gastric tube insertion (%) 212 (52.35) 155 (49.84) 57 (60.64) 0.086 Urethral catheterization (%) 203 (50.12) 142 (45.66) 61 (64.89) 0.002* Endotracheal intubation (%) 310 (76.54) 239 (76.85) 71 (75.53) 0.9 Tracheotomy(%) 40 ( 9.88) 26 ( 8.36) 14 (14.89) 0.096 Deep vein catheterization (%) 160 (39.51) 104 (33.44) 56 (59.57) < 0.001* Arterial Catheterization (%) 109 (26.91) 72 (23.15) 37 (39.36) 0.003* Drainage tube (%) 206 (50.86) 157 (50.48) 49 (52.13) 0.871 CRRT (%) 48 (11.85) 19 ( 6.11) 29 (30.85) < 0.001* Age (years) 61.00 [50.00, 72.00] 61.00 [49.00, 72.00] 61.50 [53.00, 72.00] 0.688 Temperature (℃) 37.10 [36.60, 37.80] 37.00 [36.60, 37.80] 37.30 [36.80, 38.00] 0.115 SBP ((mmHg) 126.00[116.00,138.00] 127.00 [117.00138.00] 122.00[110.00,135.00] 0.016* DBP ((mmHg) 70.00 [64.00, 78.00] 70.00 [65.00, 78.00] 70.00 [62.00, 76.00] 0.309 Heart rate (min − 1 ) 91.00 [80.00, 103.00] 90.00 [80.00, 102.00] 94.50 [82.00, 110.00] 0.069 Respiratory rate (min − 1 ) 18.00 [15.00, 20.00] 18.00 [15.00, 20.00] 18.00 [15.00, 20.00] 0.754 PT (s) 15.50 [14.40, 17.10] 15.30 [14.30, 16.75] 16.70 [15.00, 19.10] < 0.001* PT %(%) 73.00 [61.00, 84.00] 75.00 [63.00, 85.50] 64.00 [50.00, 77.75] < 0.001* INR 1.21 [1.11, 1.37] 1.19 [1.10, 1.35] 1.33 [1.16, 1.60] < 0.001* APTT (s) 39.30 [34.70, 45.30] 38.20 [34.10, 43.70] 41.70 [38.15, 48.32] < 0.001* FIB (g/L) 4.36 [2.74, 5.96] 4.34 [2.70, 6.04] 4.48 [2.94, 5.78] 0.506 TT (s) 17.00 [15.90, 18.60] 16.90 [15.90, 18.45] 17.25 [15.80, 19.08] 0.311 D-D ( µg/mL) 2.95 [1.80, 9.13] 2.85 [1.71, 8.84] 3.62 [2.03, 9.96] 0.176 FDP (µg/mL) 12.71 [6.60, 33.65] 12.10 [6.30, 31.56] 15.53 [8.00, 37.23] 0.047* WBC (10 9 /L ) 11.11 [7.85, 15.66] 11.11 [7.94, 15.09] 11.16 [7.62, 16.96] 0.615 Neu% (%) 87.90 [83.20, 91.20] 87.00 [82.70, 90.50] 89.70 [86.03, 93.65] < 0.001* Lym% (%) 6.40 [3.90, 10.00] 6.70 [4.10, 10.40] 5.30 [3.02, 8.20] 0.001* Mon% (%) 4.80 [3.10, 6.60] 5.10 [3.50, 6.80] 3.55 [2.02, 5.25] < 0.001* Eos% (%) 0.10 [0.00, 0.60] 0.10 [0.00, 0.70] 0.10 [0.00, 0.50] 0.986 Bas% (%) 0.20 [0.10, 0.30] 0.20 [0.10, 0.30] 0.20 [0.10, 0.30] 0.177 Neu (10 9 /L) 9.79 [6.81, 13.75] 9.69 [6.83, 13.07] 10.13 [6.38, 15.59] 0.337 Lym (10 9 /L) 0.69 [0.42, 1.09] 0.73 [0.44, 1.17] 0.57 [0.33, 0.79] < 0.001* Mon (10 9 /L) 0.51 [0.29, 0.83] 0.55 [0.33, 0.87] 0.41 [0.20, 0.66] 0.001* Eos (10 9 /L) 0.01 [0.00, 0.07] 0.01 [0.00, 0.07] 0.01 [0.00, 0.06] 0.666 Bas (10 9 /L) 0.02 [0.01, 0.03] 0.02 [0.01, 0.03] 0.02 [0.01, 0.04] 0.178 RBC (10 12 /L) 3.21 [2.65, 3.88] 3.20 [2.66, 3.94] 3.25 [2.62, 3.75] 0.517 Hb (g/L) 98.00 [82.00, 119.00] 98.00 [82.00, 119.50] 98.50 [81.25, 114.00] 0.405 HCT (%) 30.10 [25.00, 35.90] 30.30 [25.00, 36.55] 29.40 [24.92, 34.25] 0.336 MCV (fL) 93.50 [89.80, 97.40] 93.50 [89.80, 97.30] 93.50 [89.32, 97.68] 0.765 MCH (pg ) 30.50 [29.40, 31.80] 30.50 [29.30, 31.80] 30.50 [29.63, 31.60] 0.991 MCHC ( g/L) 327.00[316.00,337.00] 328.00[316.00,337.00] 326.00[316.50,338.00] 0.705 RDW -CV (%) 14.00 [13.20, 15.50] 13.90 [13.20, 15.35] 14.60 [13.30, 15.88] 0.01* RDW-SD (fL) 47.80 [43.80, 52.10] 47.50 [43.50, 51.85] 48.80 [45.02, 53.08] 0.115 PLT (10 9 /L) 140.00 [81.00, 207.00] 156.00[94.00,213.50] 96.50 [57.25, 159.75] < 0.001* PCT (%) 0.16 [0.10, 0.23] 0.18 [0.11, 0.24] 0.12 [0.07, 0.19] < 0.001* PDW (fL) 14.20 [12.10, 16.50] 14.00 [11.95, 16.10] 15.10 [12.80, 17.78] 0.009* MPV (fL) 11.50 [10.70, 12.50] 11.50 [10.60, 12.40] 11.80 [10.90, 12.80] 0.025* TP (g/L) 56.87 (8.62) 57.34 (8.70) 55.34 (8.20) 0.049* ALB (g/L) 34.34 (5.74) 34.74 (5.77) 33.02 (5.49) 0.011* GLB (g/L) 22.49 (5.76) 22.54 (5.83) 22.32 (5.57) 0.754 A/G 1.53 [1.31, 1.82] 1.56 [1.31, 1.83] 1.47 [1.24, 1.77] 0.24 TBIL (µmol/L) 17.20 [11.04, 30.22] 15.85 [10.27, 27.29] 24.65 [13.48, 63.41] < 0.001* DBIL (µmol/L) 5.16 [2.86, 12.46] 4.30 [2.56, 10.20] 9.71 [4.90, 37.19] < 0.001* IBIL (µmol/L) 11.28 [7.50, 17.95] 10.62 [7.32, 16.30] 14.33 [7.88, 30.06] 0.003* ALT (U/L) 32.50 [17.50, 100.70] 31.00 [16.55, 96.90] 38.90 [19.60, 146.35] 0.233 AST (U/L) 46.00 [29.20, 102.50] 47.00 [29.45, 102.20] 44.90 [28.95, 106.03] 0.97 ALP (U/L) 83.80 [63.60, 110.00] 80.20 [60.80, 104.00] 95.90 [67.45, 128.30] 0.003* GGT (U/L) 42.40 [21.20, 79.40] 39.20 [20.05, 76.05] 47.60 [29.30, 106.50] 0.054* BUN (mmol/L ) 9.50 [6.38, 14.95] 9.03 [6.14, 14.25] 12.36 [7.89, 17.67] 0.001* CRE (µmol/L) 78.60 [57.40, 131.60] 75.10 [55.25, 117.35] 90.90 [67.73, 197.33] 0.001* UA (µmol/L) 270.30[165.30,399.10] 272.80[169.30,389.15] 251.20[155.43,424.72] 0.773 eGFR(ml/(min·1.73m 2 )) 83.00 [46.00, 104.00] 86.00 [55.50, 105.50] 66.00 [26.25, 96.75] 0.002* GLU (mmol/L) 8.50 [6.46, 11.30] 8.44 [6.58, 11.12] 9.04 [6.13, 11.74] 0.868 K (mmol/L) 4.06 [3.70, 4.50] 4.05 [3.69, 4.48] 4.16 [3.72, 4.56] 0.627 Na (mmol/L) 141.10[137.60,146.50] 141.30[137.95,146.50] 140.75[136.62,146.38] 0.428 Cl (mmol/L) 105.10[101.40,110.00] 105.20[101.80,110.00] 104.25[100.70,109.95] 0.276 HCO3 (mmol/L) 24.30 [21.70, 27.10] 24.50 [22.20, 27.20] 22.90 [19.83, 26.58] 0.003* Procalcitonin (ng/mL) 1.05 [0.33, 5.58] 0.64 [0.24, 1.90] 14.13 [4.10, 33.67] < 0.001* CRP ( mg/L) 57.90[15.20,139.47] 41.32 [10.00, 99.51] 149.87 [84.61, 178.24] < 0.001* Absolute numbers and percentages are used for categorical variables and mean and standard deviation are used for continuous variables *shows the significant differences between the GN-BSI and Non-GN-BSI Variables of importance In the training set, the optimal lambda value was determined to be 0.059 using LASSO regression (Fig. 2A and B), with seven variables identified as potential predictors from the baseline characteristics. These variables are listed in detail in Supplementary Table 2.Subsequently, the seven variables selected by LASSO regression were incorporated into a multivariate logistic regression model. Of these, four variables shown in the Table 2 were retained as optimal predictive features: deep vein catheterization, continuous renal replacement therapy (CRRT), procalcitonin, and C-reactive protein (CRP). Multimodel integrated analysis for classification All seven machine learning models demonstrated excellent predictive performance in the training set(Supplementary Table 3). For the hyperparameters of each model,please refer to Supplementary Table 4 in the supplementary material. In the validation set,the ROC curves and confusion matrices of each model are shown in Fig. 3A and Fig. 4, and all models achieved accuracy and AUC values of 0.85 or above. From the evaluation metrics in Table 3 ,the XGBoost model demonstrated comprehensive superiority, achieving the highest F1-score,a metric that balances precision and sensitivity,and outperforming all other models in sensitivity, specificity, Kappa value, Youden's J statistic, and positive and negative predictive values. The XGBoost model demonstrated excellent predictive performance, with an AUC of 0.898 (95% CI: 0.807–0.959), an accuracy of 88.43%, a precision of 85.00%, and a sensitivity of 60.71%. Additionally, it achieved a specificity of 96.77%, an F1-score of 70.83%, a PPV of 85.00%, and an NPV of 89.10%.Considering the F1 score and the other evaluation metrics, the XGBoost model emerged as the best model. The Brier score of the calibration curve for the XGBoost model is 0.101, indicating that they have relatively excellent prediction performance (Fig. 3B). According to the DCA curve (Fig. 3C), the XGBoost model showed a large net benefit,indicating that the established model has robust clinical validity. Interpretability and application of the model The feature’s contribution to the prediction outcomes was quantified using SHAP, which provides more insights into how the XGBoost model predicted outcomes.The SHAP feature importance ranking depict the standard bar chart of the mean absolute SHAP value for each feature in descending order(Fig. 5A). The SHAP beeswarm plot provides insights into the predictions of the XGBoost model. The diagram illustrates the degree of influence each feature has on model output(Fig. 5B). The color gradient (feature value)from red to blue represents the magnitude of the feature value, with red denoting a high feature value and blue indicating a low feature value. A positive Shapley value for each feature indicates an increased risk of GN-BSI while a negative value suggests decreased risk. Figure 6 is SHAP dependency plot,which analyze the impact of features at factor level on the risk of the predictive model. It demonstrated that higher PCT and CRP values, performed deep vein catheterization and CRRT contributed to an elevated risk of GN-BSI in ICU patients.Additionally, the force plots provided personalized feature attributions using two representative examples and illustrated how to use SHAP to interpret the predictions of individual model, as shown in Fig. 7A (a ICU patient without GN-BSI) and B (a ICU patient with GN-BSI). Figure 1.A flowchart describing the number of patients included in the analysis after exclusion criteria. The ICU patients included were divided into those with and without GN-BSI Figure 2.Presentation of the results of the LASSO regression analysis. (A) LASSO Regression Model Factor Selection: Left line represents the optimal lambda value (lambda⋅min), while the right line marks the lambda value within one standard error of the optimal (lambda.1se); (B) LASSO regression model screening variable trajectories. Figure 3.The performance and comparison of seven different predictive models, (A) ROC curves and AUC values of the validation set (B) Calibration curves of the XGBoost in the validation set (C) Decision curves analysis of the XGBoost in the validation set. Figure 4.Confusion Matrices of ML models. (A) Logistic Regression’s Confusion Matrices. (B) Decision Tree’s Confusion Matrices. (C) Random Forest’s Confusion Matrices. (D) XGBoost’s Confusion Matrices. (E) LightGBM’s Confusion Matrices. (F) SVM’s Confusion Matrices. (G) ANN’s Confusion Matrices. Table 2 Multivariate logistic regression analysis of GN-BSI in ICU patients Variables β SE Wald OR 95%CI P- value Carbapenem antibiotics 0.507 0.454 1.117 1.66 0.682–4.037 0.264 Deep vein catheterization 0.903 0.428 2.108 2.466 1.065–5.708 0.035* CRRT 1.336 0.579 2.309 3.806 1.224–11.834 0.021* APTT 0.03 0.016 1.84 1.03 0.998–1.064 0.066 DBIL 0.006 0.004 1.385 1.006 0.998–1.014 0.166 Procalcitonin 0.074 0.018 4.163 1.077 1.04–1.115 < 0.001* CRP 0.009 0.003 2.743 1.009 1.003–1.016 0.006* *shows the significant differences between the GN-BSI and Non-GN-BSI Table 3 Evaluation metrics results of seven models in the validation set Model AUC(95%CI) Accuracy Sensitivity Specificity F1 Score Kappa Youden's J PPV NPV Logistic regression 0.889 (0.807–0.955) 0.851 0.5 0.957 0.609 0.522 0.457 0.778 0.864 Decision Tree 0.877 (0.790–0.946) 0.876 0.643 0.946 0.706 0.628 0.589 0.783 0.898 Random Forest 0.874 (0.790–0.941) 0.843 0.536 0.935 0.612 0.516 0.471 0.714 0.87 XGBoost 0.898 (0.807–0.959) 0.884 0.607 0.968 0.708 0.639 0.575 0.85 0.891 LightGBM 0.872 (0.782–0.941) 0.826 0.571 0.903 0.604 0.493 0.475 0.64 0.875 SVM 0.910 (0.826–0.971) 0.868 0.5 0.978 0.636 0.563 0.478 0.875 0.867 ANN 0.911 (0.834–0.968) 0.876 0.643 0.946 0.706 0.628 0.589 0.783 0.898 Figure 5.Feature importance analysis by SHAP method (A) The SHAP feature importance ranking plot of features of the XGBoost model. (B) SHAP beeswarm plot of the XGBoost model. Figure 6.SHAP dependency plot of features in the XGBoost model. The Y-axis represents SHAP values, while the X-axis represents actual clinical parameters.For binary variables such as Deep vein catheterization and CRRT, “0” indicates the absence of the condition, while “1” indicates its presence.. Significantly, when a feature’s SHAP value is greater than 0, it suggests an increased risk of GN-BSI, whereas a negative SHAP value suggests a reduced risk. Figure 7.SHAP force plot of features in the XGBoost model. The force plots provide personalized feature attributions using two representative examples. A: a ICU patient without GN-BSI; B: a ICU patient with GN-BSI. Discussion Constructing a predictive model for Gram-negative bacterial bloodstream infection (GNB-BSI) in the ICU is critical for the early identification of pathogens, improving patient survival rates, and combating antimicrobial resistance. To this end, this study developed and compared several prediction models using a range of machine learning algorithms. Among them, the eXtreme Gradient Boosting (XGBoost) model demonstrated the most favorable predictive value, confirming its utility as an effective tool for early clinical intervention and for delaying the development of bacterial resistance in this high-risk patient population. As far as we know, Feature selection is a crucial process in the development of predictive models. In this study, Lasso regression was first employed to identify seven key predictors. Following this, multivariate logistic regression was subsequently applied to refine the selection, ultimately identifying four final variables, namely deep vein catheterization, CRRT, Procalcitonin, and CRP. The relationship between these features and BSI has been extensively examined in prior research. Regarding deep vein catheterization, it is a significant risk factor for hospital-acquired bloodstream infections[ 22 ]. The catheter, as a foreign body, serves as a conduit for skin flora (both resident and transient) to enter the bloodstream. Simultaneously, a biofilm rapidly forms on its surface, providing a niche for bacterial colonization. This biofilm confers antibiotic resistance and is a common cause of recurrent bacteremia. Among high-risk populations such as intensive care and hemodialysis patients, the incidence of catheter-related bloodstream infections (CRBSI) ranges from approximately 0.5 to 5.0 per 1000 catheter-days [ 23 ], with Gram-negative bacteria, including Escherichia coli and Klebsiella species, accounting for approximately 30% of the causative pathogens[ 24 ].A biological link has been established between CRRT in ICU and GNB-BSI, with research indicating that patients requiring CRRT are typically critically ill, immunocompromised individuals in the ICU, which predisposes them to bacterial infections[ 25 ]. Their heightened susceptibility to bloodstream infections is largely attributable to the prolonged use of central venous catheters, resulting in a significantly greater risk compared to non-CRRT patients or general nursing populations. While Gram-positive bacteria remain the primary pathogens, the incidence of Gram-negative infections has risen notably in recent years[ 26 ]. Gatti M et al[ 27 ]. also pointed out that end-stage renal disease and central catheterization are independent risk factors for GN-BSI. Previous studies have shown that PCT and CRP are potential biomarkers for BSI[ 28 ]. In healthy individuals,PCT levels are typically undetectable or very low (< 0.05 ng/mL). However, during severe systemic bacterial infections like sepsis or bacteremia, endotoxins and inflammatory cytokines (including TNF-α, IL-6) trigger a substantial release of PCT from iver macrophages and systemic tissue cells throughout the body, resulting in a marked elevation of serum concentrations[ 29 ]. Furthermore, PCT levels are typically more elevated in Gram-negative bloodstream infections[ 30 ], suggesting their potential utility in predicting the causative pathogen type. CRP is an acute-phase reactant whose levels rise rapidly in response to a wide range of conditions, including inflammation, infection (bacterial, viral, or fungal), tissue damage, and malignancy. Consequently, it lacks the specificity required for diagnosing bloodstream infections, and its diagnostic performance in this context is inferior to that of PCT[ 31 ]. Therefore, the combined application of PCT and CRP plays a synergistic role in significantly enhancing the diagnostic accuracy for bloodstream infections and facilitating earlier clinical intervention[ 32 ]. Our results indicate that the XGBoost model demonstrated superior overall performance compared to the other six machine learning algorithms across key metrics, including accuracy, sensitivity, specificity, AUROC values and F1-score in the validation set, as did XGBoost in the training set. According to the DCA of the training and validation sets, the intervention measures guided by the predictive model produced outstanding results, except for a small range of low preferences. Furthermore, the model calibration was evaluated using a calibration curve, and the Brier score of XGBoost in the validation set was 0.102 (95% CI 0.070–0.137), demonstrating strong calibration and prediction accuracy. Based on its superior discriminative ability, well-calibrated probabilistic outputs, and significant clinical net benefit demonstrated through comprehensive evaluation, the XGBoost algorithm was selected as the optimal predictive tool for Gram-negative bacteremia in ICU patients. Unlike previous models primarily designed for predicting generalized bloodstream infections, this study establishes a specialized framework for the early detection of GNB-BSI[ 33 ]. Given that GNB-BSI is frequently associated with rapid clinical deterioration, a heightened risk of septic shock, and significant antimicrobial resistance, the capacity for early and precise pathogen orientation offers critical decision support. This enables clinicians to initiate timely and targeted anti-Gram-negative therapy, thereby enhancing the model's practical utility in managing high-risk ICU patients.Moreover,this study comprehensively evaluated 81 variables encompassing demographic characteristics, relatively comprehensive set of clinical and laboratory parameters, ensuring extensive coverage of potential predictors for GNB-BSI in ICU patients.Compared to prior models requiring more variables,such as Hu et al. (5 variables) and Zhang et al. (32 variables),our model achieves a superior predictive power (XGBoost AUC: 0.898) using only 4 routine ICU indicators[ 12 , 33 ]. Our model also has its limitations. Firstly, the generalizability of this study may be limited by its relatively small sample size, which carries a potential for selection bias. Secondly, although we have made every effort to collect potential predictive factors for GNB-BSI, some important risk factors, such as a large number of missing blood gas analysis indicators, must be excluded as related variables. These may lead us to overlook some features. Finally, the model demonstrated robust performance during internal validation; however, its generalizability requires confirmation through external validation. For clinical application, we recommend that future studies adopt a consistent analytical framework aligned with our research objectives. The predicted outcomes should be interpreted within the clinical context to ensure they contribute meaningfully to improving patient prognosis. Abbreviations BSI Bloodstream Infection GN-BSI Gram-negative Bloodstream Infection ICU Intensive Care Unit ML Machine learning AUC Area under the receiver operating characteristic curve SHAP Shapley Additive Explanation LASSO L1-penalty least absolute shrinkage and selection operator RF Random Forest XGBoost eXtreme Gradient Boosting LightGBM Light Gradient Boosting Machine SVM Support Vector Machine ANN Artificial Neural Network PPV Positive predictive value NPV Negative predictive value DCA Decision curve analysis CRRT Continuous renal replacement therapy SBP Systolic blood pressure DBP Diasto blood licpressure PT Prothrombin Time PT% Prothrombin Activity INR International Normalized Ratio APTT Activated Partial Thromboplastin Time FIB Fibrinogen TT Thrombin Time D-D D-dimer FDP Fibrinogen Degradation Products WBC White Blood Cell Count Neu% Neutrophils Percentage Lym% Lymphocyte Percentage Mon% Monocytes Percentage Eos% Eosinophils Percentage Bas% Basophils Percentage Neu Neutrophil Count Lym Lymphocyte Count Mon Monocyte Count Eos Direct Eosinophil Count Bas Direct Basophil Count RBC Red Blood Cell Count Hb Hemoglobin HCT Hematocrit MCV Mean Corpuscular Volume MCH Mean Corpuscular Hemoglobin MCHC Mean Corpuscular Hemoglobin Concentration RDW-CV Red Blood Cell Distribution Width - Coefficient of Variation RDW-SD Red Blood Cell Distribution Width - Standard Deviation PLT Platelet Count PCT Plateletcrit PDW Platelet Distribution Width MPV Mean Platelet Volume TP Total Protein ALB Albumin GLB Globulin A/G Alb/Glb Ratio TBIL Total Bilirubin DBIL Direct Bilirubin IBIL Indirect Bilirubin ALT Alanine Aminotransferase AST Aspartate Aminotransferase ALP Alkaline Phosphatase GGT Gamma-glutamyl Transferase BUN Blood Urea Nitrogen CRE Creatinine UA Uric Acid eGFR Estimated Glomerular Filtration Rate GLU Glucose K Potassium Na Sodium Cl Chlorine HCO3 Bicarbonate CRP C-Reactive Protein PH Potential of Hydrogen PO2 Partial Pressure of Oxygen PCO2 Partial Pressure of Carbon Dioxide BE Partial Pressure of Carbon Dioxide Lac Lactic Acid Declarations Ethics approval and consent to participate The study protocol received approval from the Clinical Medical Research Ethics Committee of the First Affiliated Hospital of Anhui Medical University (PJ2025-11-09) and was conducted in accordance with the principles of the Declaration of Helsinki. Due to the retrospective nature of the study, the requirement for written informed consent was waived. All patient data were de-identified prior to analysis to ensure confidentiality. Consent for the publication Not applicable. Competing interests The authors declare no competing interests. Authors' information 1 Department of Clinical Laboratory, The First Affiliated Hospital of Anhui Medical University, Hefei, Anhui Province, China 2 Department of Reproductive Medicine, The First Affiliated Hospital of Anhui Medical University, Hefei, Anhui Province, China Funding This work was supported by the First Affiliated Hospital of Anhui Medical University Horizontal Research Project(CXPJJH21002-2021SS) Author Contribution Research design and conception were performed by Ya-Ling Zhou, Zhong-Le Cheng, and Zhong-Xin Wang. Data collection and statistical analysis were carried out by Ya-Ling Zhou，Hong-ting Da and Ting-Ting Wang. Ya-Ling Zhou developed the prediction system based on the proposed model. The initial draft of the manuscript was written by Ya-Ling Zhou, and Zhong-Le Cheng critically revised it for important intellectual content. All authors have reviewed and approved the final version of the manuscript. Acknowledgements Not applicable. Data Availability The datasets used and analyzed in this study are available from the corresponding author upon reasonable request. References Zheng C, Chen Q, Pan S, Li Y, Zhong L, Zhang X, Cui W, Lin R, Zhang G, Zhang S. Staphylococcus aureus bloodstream infection in a Chinese tertiary-care hospital: A single-center retrospective study. Chin Med J (Engl). 2023;136(12):1503–5. Kern WV, Rieg S. Burden of bacterial bloodstream infection-a brief update on epidemiology and significance of multidrug-resistant pathogens. Clin Microbiol Infect. 2020;26(2):151–7. Pradubkham T, Suwanpimolkul G, Gross AE, Nakaranurack C. Intravenous to oral transition of antibiotics for gram-negative bloodstream infection at a University hospital in Thailand: Clinical outcomes and predictors of treatment failure. PLoS ONE. 2022;17(9):e0273369. Al-Hasan MN. Gram-Negative Bloodstream Infection: Implications of Antimicrobial Resistance on Clinical Outcomes and Therapy. Antibiot (Basel). 2020;9(12):922. Scaglione G, Perego M, Colaneri M, Genovese C, Brivio F, Covizzi A, Viaggi B, Bandera A, Gori A, Finazzi S, Palomba E. Understanding the burden of antibiotic resistance: a decade of carbapenem-resistant Gram-negative bacterial infections in Italian intensive care units. Front Microbiol. 2024;15:1405390. Kundar R, Gokarn K, CRISPR-Cas System. A Tool to Eliminate Drug-Resistant Gram-Negative Bacteria. Pharmaceuticals (Basel). 2022;15(12):1498. Gu S, Xiong J, Peng S, Hu L, Zhu H, Xiao Y, Luo H, Hang Y, Chen Y, Fang X, Cao X, Fang Y, Li F, Zhu J, Zhong Q. Assessment of Effective Antimicrobial Regimens and Mortality-Related Risk Factors for Bloodstream Infections Caused by Carbapenem-Resistant Acinetobacter baumannii . Infect Drug Resist. 2023;16:2589–600. Warren BG, Yarrington ME, Polage CR, Anderson DJ, Moehring RW. Evaluation of hospital blood culture utilization rates to identify opportunities for diagnostic stewardship. Infect Control Hosp Epidemiol. 2023;44(2):200–5. Timsit JF, Ruppé E, Barbier F, Tabah A, Bassetti M. Bloodstream infections in critically ill patients: an expert statement. Intensive Care Med. 2020;46(2):266–84. Lynch CJ, Liston C. New machine-learning technologies for computer-aided diagnosis. Nat Med. 2018;24(9):1304–5. Richens JG, Lee CM, Johri S. Author Correction: Improving the accuracy of medical diagnosis with causal machine learning. Nat Commun. 2021;12(1):2108. Zhang F, Wang H, Liu L, Su T, Ji B. Machine learning model for the prediction of gram-positive and gram-negative bacterial bloodstream infection based on routine laboratory parameters. BMC Infect Dis. 2023;23(1):675. Hu X, Zhi S, Li Y, Cheng Y, Fan H, Li H, Meng Z, Xie J, Tang S, Li W. Development and application of an early prediction model for risk of bloodstream infection based on real-world study. BMC Med Inf Decis Mak. 2025;25(1):186. Roimi M, Neuberger A, Shrot A, Paul M, Geffen Y, Bar-Lavie Y. Early diagnosis of bloodstream infections in the intensive care unit using machine-learning algorithms. Intensive Care Med. 2020;46(3):454–62. Chicco D, Oneto L, Tavazzi E. Eleven quick tips for data cleaning and feature engineering. PLoS Comput Biol. 2022;18(12):e1010718. Maletzky A, Böck C, Tschoellitsch T, Roland T, Ludwig H, Thumfart S, Giretzlehner M, Hochreiter S, Meier J. Lifting Hospital Electronic Health Record Data Treasures: Challenges and Opportunities. JMIR Med Inf. 2022;10(10):e38557. Ma C, Sun G, Yang X, Yang S. A clinically applicable prediction model for the risk of in-hospital mortality in solid cancer patients admitted to intensive care units with sepsis. J Cancer Res Clin Oncol. 2023;149(10):7175–85. Faquih T, van Smeden M, Luo J, le Cessie S, Kastenmüller G, Krumsiek J, Noordam R, van Heemst D, Rosendaal FR, van Hylckama Vlieg A, van Willems K, Mook-Kanamori DO. A Workflow for Missing Values Imputation of Untargeted Metabolomics Data. Metabolites. 2020;10(12):486. Sun K, Huang SH, Wong DSH, Jang SS. Design and application of a variable selection method for multilayer perceptron neural network with LASSO. IEEE Trans Neural Netw Learn Syst. 2017;28:1386–96. Feng S, Huang S, Zhou Z, Zhang B, Huang C, Chen T, Zhou C, Wu S, Zhu J, Chen J, Xue J, Zhan X, Liu C. Development and Validation of a Machine Learning-Based Online Prognostic Model for Cervical Spondylosis Patients After Anterior Cervical Discectomy and Fusion: A Multicenter Study. JOR Spine. 2025;8(3):e70090. Guo Z, Wang T, Chi S, Huang L. Energy consumption analysis and prediction in exercise training based on accelerometer sensors and deep learning. Sci Rep. 2025;15(1):19423. Abubakar U. Point-prevalence survey of hospital acquired infections in three acute care hospitals in Northern Nigeria. Antimicrob Resist Infect Control. 2020;9(1):63. Pitiriga V, Bakalis J, Theodoridou K, Kanellopoulos P, Saroglou G, Tsakris A. Lower risk of bloodstream infections for peripherally inserted central catheters compared to central venous catheters in critically ill patients. Antimicrob Resist Infect Control. 2022;11(1):137. Neoh KG, Li M, Kang ET, et al. Surface modification strategies for combating catheter-related complications: recent advances and challenges[J]. J Mater Chem B. 2017;5(11):2045–67. Brunelli SM, Turenne W, Sibbel S, et al. Clinical and economic burden of bloodstream infections in critical care patients with central venous catheters[J]. J Crit Care. 2016;35:69–74. Cetin S, Dokmetas I, Hamidi AA, Bayraktar B, Gunduz A, Sevgi DY. Comparison of Risk Factors and Outcomes in Carbapenem-Resistant and Carbapenem-Susceptible Gram-Negative Bacteremia. Sisli Etfal Hastan Tip Bul. 2021;55(3):398–404. Gatti M, Bonazzetti C, Tazza B, Pascale R, Miani B, Malosso M, Beci G, Marzolla D, Rinaldi M, Viale P, Giannella M. Impact on clinical outcome of follow-up blood cultures and risk factors for persistent bacteraemia in patients with gram-negative bloodstream infections: a systematic review with meta-analysis. Clin Microbiol Infect. 2023;29(9):1150–8. Stoller J, Halpin L, Weis M, Aplin B, Qu W, Georgescu C, et al. Epidemiology of severe sepsis: 2008–2012. J Crit Care. 2016;31(1):58–62. Shim YE, Shin S, Ko Y, Kim DH, Lim SJ, Jung JH, et al. Serum procalcitonin as a biomarker for differentiating between infectious and non-infectious Fever after pancreas transplantation. Clin Transpl. 2021;35(4):e14224. Yang X, Zeng J, Yu X, Wang Z, Wang D, Zhou Q, Bai T, Xu Y, PCT. IL-6, and IL-10 facilitate early diagnosis and pathogen classifications in bloodstream infection. Ann Clin Microbiol Antimicrob. 2023;22(1):103. Shokripour M, Omidifar N, Salami K, Moghadami M, Samizadeh B. Diagnostic Accuracy of Immunologic Biomarkers for Accurate Diagnosis of Bloodstream Infection in Patients with Malignancy: Procalcitonin in Comparison with C-Reactive Protein. Can J Infect Dis Med Microbiol. 2020;2020:8362109. Liang P, Yu F. Value of CRP, PCT, and NLR in Prediction of Severity and Prognosis of Patients With Bloodstream Infections and Sepsis. Front Surg. 2022;9:857218. Hu X, Zhi S, Li Y, Cheng Y, Fan H, Li H, Meng Z, Xie J, Tang S, Li W. Development and application of an early prediction model for risk of bloodstream infection based on real-world study. BMC Med Inf Decis Mak. 2025;25(1):186. Additional Declarations No competing interests reported. Supplementary Files SupplementaryMaterial.docx Cite Share Download PDF Status: Published Journal Publication published 31 Jan, 2026 Read the published version in BMC Medical Informatics and Decision Making → Version 1 posted Editorial decision: Revision requested 14 Jan, 2026 Reviews received at journal 02 Jan, 2026 Reviewers agreed at journal 06 Dec, 2025 Reviews received at journal 06 Dec, 2025 Reviewers agreed at journal 04 Dec, 2025 Reviewers agreed at journal 04 Dec, 2025 Reviewers invited by journal 04 Dec, 2025 Editor invited by journal 04 Dec, 2025 Editor assigned by journal 03 Dec, 2025 Submission checks completed at journal 03 Dec, 2025 First submitted to journal 30 Nov, 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-8240564","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":556305211,"identity":"900c8a65-3886-48dc-b501-333189f02c16","order_by":0,"name":"Ya-Ling Zhou","email":"","orcid":"","institution":"First Affiliated Hospital of Anhui Medical University","correspondingAuthor":false,"prefix":"","firstName":"Ya-Ling","middleName":"","lastName":"Zhou","suffix":""},{"id":556305212,"identity":"16097394-f76b-4c90-803f-96344e018168","order_by":1,"name":"Hong-Ting Da","email":"","orcid":"","institution":"First Affiliated Hospital of Anhui Medical University","correspondingAuthor":false,"prefix":"","firstName":"Hong-Ting","middleName":"","lastName":"Da","suffix":""},{"id":556305214,"identity":"d3785706-86fc-4ce6-9230-044062e8ee3d","order_by":2,"name":"Ting-Ting Wang","email":"","orcid":"","institution":"First Affiliated Hospital of Anhui Medical University","correspondingAuthor":false,"prefix":"","firstName":"Ting-Ting","middleName":"","lastName":"Wang","suffix":""},{"id":556305215,"identity":"0c806a89-0fe1-480e-bc66-6a398deeb0ee","order_by":3,"name":"Zhong-Xin Wang","email":"","orcid":"","institution":"First Affiliated Hospital of Anhui Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zhong-Xin","middleName":"","lastName":"Wang","suffix":""},{"id":556305217,"identity":"7b1f732b-9080-4855-a423-679386f51b4a","order_by":4,"name":"Zhong-Le Cheng","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAo0lEQVRIiWNgGAWjYDACCSB+UCFhx0+aloQzFsmSDSRpSWyrYNxwgFgd/LObH39InCfBbHCA+eGjG0RZcueYgUHiNgk+yQNsxsY5RFlzI4chAaiFme8AD5s0UVrkgVoOJM6RYGwgWovBjRzGhsQGCcYJRGsxvJFmzJBwTCJZsplYv8jdSH784UNNnR0/e/PDx8R5Hw6YSVM+CkbBKBgFowAfAADHUy7kbJCp/wAAAABJRU5ErkJggg==","orcid":"","institution":"First Affiliated Hospital of Anhui Medical University","correspondingAuthor":true,"prefix":"","firstName":"Zhong-Le","middleName":"","lastName":"Cheng","suffix":""}],"badges":[],"createdAt":"2025-11-30 07:08:16","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8240564/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8240564/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12911-026-03361-z","type":"published","date":"2026-01-31T15:58:11+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":97720535,"identity":"df321c53-2481-4460-a588-f9999a04b6f9","added_by":"auto","created_at":"2025-12-08 15:35:44","extension":"png","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":58476,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.1.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/34e2df5fefd8dcfe15f3cd9c.png"},{"id":97720536,"identity":"1a40ecdc-1d8f-4c1e-af00-84ea65b1ea6f","added_by":"auto","created_at":"2025-12-08 15:35:44","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":62229,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/e9f8a15923304f80b1a8bc0a.docx"},{"id":97894683,"identity":"c5838947-19b5-4913-8848-6c307c527fb3","added_by":"auto","created_at":"2025-12-10 15:32:53","extension":"png","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":616003,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.2.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/68f1b92d9e58c245c75021ec.png"},{"id":97895181,"identity":"50e6f674-ccaf-415a-9a66-a5fb0bb1de7e","added_by":"auto","created_at":"2025-12-10 15:33:42","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1263378,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.3.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/de6fcf17903eee1396851633.png"},{"id":97895741,"identity":"1d1a0306-fb27-4bb5-8eae-4f95bd7ac7da","added_by":"auto","created_at":"2025-12-10 15:34:50","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1002027,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.4.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/63a309c1cc312819c6332ee1.png"},{"id":97720549,"identity":"0a9e1645-c3f2-47c9-8fa3-84623fcf920f","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":399014,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.5.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/e3a38750b6524dff329fb746.png"},{"id":97894744,"identity":"7db09925-0768-462d-aab7-f908dff3c0e6","added_by":"auto","created_at":"2025-12-10 15:32:58","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":334359,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.6.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/a4ae6bb3bf64be85d778ffb9.png"},{"id":97720554,"identity":"703f3d91-d05a-4520-b5ea-f9655882b096","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":155214,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.7.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/7935bc9a260b810d6abc49a6.png"},{"id":97895530,"identity":"5d239754-47eb-4380-b98b-59ac705c1390","added_by":"auto","created_at":"2025-12-10 15:34:24","extension":"json","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":8003,"visible":true,"origin":"","legend":"","description":"","filename":"926c849494124d6b9c92d9f12b86aa0c.json","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/df341bcc7d335a9fa0fa7be0.json"},{"id":97720546,"identity":"a9238f99-bf52-4530-9f6d-e79242b79ee2","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"xlsx","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13329,"visible":true,"origin":"","legend":"","description":"","filename":"Listofabbreviations.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/ddb20ab65422b23031b19be0.xlsx"},{"id":97720559,"identity":"632cfdf7-0cfb-4e50-9db5-72c023a1397b","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"docx","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":81089,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/c16411438d01dc04c8037b27.docx"},{"id":97896136,"identity":"cceb10a5-4f53-47be-82e5-3bb1e55a9b85","added_by":"auto","created_at":"2025-12-10 15:35:57","extension":"xml","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":146499,"visible":true,"origin":"","legend":"","description":"","filename":"926c849494124d6b9c92d9f12b86aa0c1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/bb06184877e374724380f372.xml"},{"id":97720557,"identity":"65fa78c1-8733-4ec4-918f-61ecdcb7259c","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":58476,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.1.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/8f374eb5f21462c004ea16d2.png"},{"id":97895760,"identity":"6207ef96-835e-4520-b6fd-5a91f905e4b5","added_by":"auto","created_at":"2025-12-10 15:34:52","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":616003,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.2.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/f94ca685473a3621996f52a6.png"},{"id":97894900,"identity":"3b5ec7da-3f17-4303-88d7-d6ec9b8c416a","added_by":"auto","created_at":"2025-12-10 15:33:11","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1263378,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.3.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/a26ab7de9fd5aa720f83738d.png"},{"id":97894001,"identity":"eb5573c3-03cd-4153-8386-b4007be1183e","added_by":"auto","created_at":"2025-12-10 15:31:47","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1002027,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.4.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/b9b0db6041a9109009006d3d.png"},{"id":97896138,"identity":"83152b53-bce0-4744-9926-d623459fedcc","added_by":"auto","created_at":"2025-12-10 15:35:57","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":399014,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.5.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/98f374dbca855abdf66ab922.png"},{"id":97895159,"identity":"536925e5-aaeb-42f2-b491-0c67d61cc2f6","added_by":"auto","created_at":"2025-12-10 15:33:39","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":334359,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.6.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/fd5899a11d83f86fda28cdfa.png"},{"id":97720561,"identity":"8d38a05e-8485-45d4-b47c-f81ec1e808d5","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":155214,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.7.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/387750bf4255008aeb35695b.png"},{"id":97720556,"identity":"44888d8f-c1df-402f-8145-d98b47321f78","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":25961,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig.1.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/70a9497f53334111f091b499.png"},{"id":97720569,"identity":"ca48adb0-3819-49cd-901a-417d414c05b3","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"png","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":176925,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig.2.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/a26f59ea3776f4327c727c8f.png"},{"id":97720563,"identity":"bee7800b-12c0-480c-83dd-6e86c135301e","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"png","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":298240,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig.3.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/950b3f18c5b836a85e70b472.png"},{"id":97893848,"identity":"3b1ce2c4-4e06-46bb-a5e4-30d577eeabc4","added_by":"auto","created_at":"2025-12-10 15:31:21","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":277597,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig.4.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/94a51004fcb4c044b45756fb.png"},{"id":97720565,"identity":"3831ee2d-53be-4d91-810e-d91b6d57b6e7","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"png","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":80392,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig.5.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/4f2c0b6205bf7e05e7562e1f.png"},{"id":97720571,"identity":"252f0826-fc22-4b63-9624-bc23723df87d","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"png","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":233723,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig.6.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/8222be83fb43ac60f455091b.png"},{"id":97893502,"identity":"1710f756-0f66-4114-b411-8f8f62dd1710","added_by":"auto","created_at":"2025-12-10 15:30:37","extension":"png","order_by":25,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":34670,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig.7.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/3771028c43dd3807d1a684ee.png"},{"id":97720570,"identity":"c2a6f245-0397-4259-8d7e-a27660799070","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"xml","order_by":26,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":140787,"visible":true,"origin":"","legend":"","description":"","filename":"926c849494124d6b9c92d9f12b86aa0c1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/9160c98f4ca4c4a1f7f73077.xml"},{"id":97720567,"identity":"f0011489-0bad-46e4-8291-b0876bf95962","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"html","order_by":27,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":150338,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/ed8b2d125346cfe0f61337ab.html"},{"id":97720537,"identity":"03929689-11f1-45c6-a723-3b8e2983bd99","added_by":"auto","created_at":"2025-12-08 15:35:44","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":470028,"visible":true,"origin":"","legend":"\u003cp\u003eA flowchart describing the number of patients included in the analysis after exclusion criteria. The ICU patients included were divided into those with and without GN-BSI\u003c/p\u003e","description":"","filename":"Fig.1.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/724143778af2707298203115.png"},{"id":97895611,"identity":"ff19a2be-e012-4bcc-8015-78379fa24192","added_by":"auto","created_at":"2025-12-10 15:34:33","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":93258,"visible":true,"origin":"","legend":"\u003cp\u003ePresentation of the results of the LASSO regression analysis. (A) LASSO Regression Model Factor Selection: Left line represents the optimal lambda value (lambda⋅min), while the right line marks the lambda value within one standard error of the optimal (lambda.1se); (B) LASSO regression model\u003c/p\u003e\n\u003cp\u003escreening variable trajectories.\u003c/p\u003e","description":"","filename":"Fig.2.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/b3f3b1066baac4d5a5d5dfb7.png"},{"id":97895853,"identity":"a0fc7535-ad79-437a-8751-e4a3da788f04","added_by":"auto","created_at":"2025-12-10 15:35:11","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":2925492,"visible":true,"origin":"","legend":"\u003cp\u003eThe performance and comparison of seven different predictive models, (A) ROC curves and AUC values of the validation set (B) Calibration curves of the XGBoost in the validation set (C) Decision curves analysis of the XGBoost in the validation set.\u003c/p\u003e","description":"","filename":"Fig.3.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/5c7fa4c9b253b3a488effa80.png"},{"id":97720539,"identity":"66c35304-bbf7-431e-9307-721d43ffd90b","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":219635,"visible":true,"origin":"","legend":"\u003cp\u003eConfusion Matrices of ML models. (A) Logistic Regression’s Confusion Matrices. (B) Decision Tree’s Confusion Matrices. (C) Random Forest’s Confusion Matrices. (D) XGBoost’s Confusion Matrices. (E) LightGBM’s Confusion Matrices. (F) SVM’s Confusion Matrices. (G) ANN’s Confusion Matrices.\u003c/p\u003e","description":"","filename":"Fig.4.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/3da7ca1093770af07eeb3939.png"},{"id":97720543,"identity":"74db319b-dec7-4322-a228-5b76343bc502","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":850705,"visible":true,"origin":"","legend":"\u003cp\u003eFeature importance analysis by SHAP method (A) The SHAP feature importance ranking plot of features of the XGBoost model. (B) SHAP beeswarm plot of the XGBoost model.\u003c/p\u003e","description":"","filename":"Fig.5.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/5c561c0cbf48e19161c08839.png"},{"id":97895566,"identity":"699ccadf-e430-4f4c-9a5a-6a51f57a1065","added_by":"auto","created_at":"2025-12-10 15:34:27","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":188818,"visible":true,"origin":"","legend":"\u003cp\u003eSHAP dependency plot of features in the XGBoost model. The Y-axis represents SHAP values, while the X-axis represents actual clinical parameters.For binary variables such as Deep vein catheterization and CRRT, “0”indicates the absence of the condition, while “1”indicates its presence. . Significantly, when a feature’s SHAP value is greater than 0, it suggests an increased risk of GN-BSI, whereas a negative SHAP value suggests a reduced risk.\u003c/p\u003e","description":"","filename":"Fig.6.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/5bc53d6b23f2af760a30b924.png"},{"id":97893712,"identity":"421a4f23-23b8-4b28-a841-a9eeadedd4ba","added_by":"auto","created_at":"2025-12-10 15:30:55","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":380697,"visible":true,"origin":"","legend":"\u003cp\u003eSHAP force plot of features in the XGBoost model. The force plots provide personalized feature attributions using two representative examples. A: a ICU patient without GN-BSI; B: a ICU patient with GN-BSI.\u003c/p\u003e","description":"","filename":"Fig.7.png","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/3ca41c36ea33c6ddaaebfe3d.png"},{"id":101690439,"identity":"66601568-c2d4-4f18-be70-e0c2632656da","added_by":"auto","created_at":"2026-02-02 16:02:49","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":6290109,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/36115474-e0d7-467d-acde-74fe69b8ea90.pdf"},{"id":97720538,"identity":"746a67f4-46a6-47aa-a3c8-592d58b19e2e","added_by":"auto","created_at":"2025-12-08 15:35:45","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":81089,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-8240564/v1/c5a45a1427f8c74a170bded0.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"A machine learning model for the early prediction of Gram-negative bloodstream infection in ICU patients","fulltext":[{"header":"Introduction","content":"\u003cp\u003eBloodstream infection (BSI) is a life-threatening systemic condition and a major global health challenge associated with high incidence and mortality[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. In developed countries, the incidence of BSI is estimated at 100 to 200 cases per 100,000 person-years and continues to rise[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Gram-negative bloodstream infection (GN-BSI) has attracted considerable clinical and public health concern due to its high incidence, potential to progress to severe sepsis, considerable mortality, and the escalating challenge of antimicrobial resistance[\u003cspan additionalcitationids=\"CR4\" citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. In healthcare-associated infections, particularly within the intensive care unit (ICU), Gram-negative bloodstream infections consistently account for 25%\u0026ndash;30% of cases, posing a major challenge for both infection control and clinical management[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. The crude mortality of ICU patients suffering from BSI is above 30%[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Early empirical antibiotic therapy is critical for patient survival. However, it may contribute to the development of antimicrobial resistance. Therefore, the rapid identification of pathogens is essential, as it enables clinicians to make timely and targeted adjustments to the antibiotic regimen. Currently, blood culture remains the gold standard for diagnosing bloodstream infections. Nevertheless, its time-consuming nature and high false-negative rate can delay early diagnosis and treatment[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Although rapid alternatives such as molecular detection techniques can guide timely antimicrobial therapy, their high cost and technical demands make them impractical for routine adoption[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Machine learning (ML) has demonstrated considerable potential in supporting disease diagnosis[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. In recent years, advances in ML have led to its rapid adoption across various medical disciplines. A number of ML-based models have been successfully developed, with studies confirming their feasibility and interpretability for predicting bacteremia[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. However, most existing machine learning models for bacteremia prediction rely on a broad set of conventional laboratory or clinical parameters.and some models incorporated a large number of features complicate the model and hinder clinical adoption. On this basis, we integrated the relatively complete clinical and laboratory parameters of patients to further improve the accuracy of model prediction[\u003cspan additionalcitationids=\"CR13\" citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. The purpose of this study is to find the optimal combination of these features finally, a prediction model that can early predict GN-BSI in ICU patients is developed by using machine learning algorithm, in order to provide reference for early clinical diagnosis and treatment.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eStudy population\u003c/h2\u003e\u003cp\u003eThis study employed a secondary analysis of a retrospective cohort conducted between January and July 2025 in the ICU of the West District, The First Affiliated Hospital of Anhui Medical University. The inclusion criteria were: (1) age\u0026thinsp;\u0026ge;\u0026thinsp;18 years; (2) admission to the ICU; (3) an anticipated ICU length of stay\u0026thinsp;\u0026gt;\u0026thinsp;48 hours; and (4) at least one blood culture obtained during hospitalization. The exclusion criteria were: (1) a blood culture positive for non-gram-negative bacteria; (2) a concurrent non-Gram-negative bacterial infection identified within 7 days before or after the index Gram-negative bacteremia episode; and (3) pregnancy or lactation. Clinical and laboratory data pertaining to Gram-negative bacteremia were collected for all enrolled adult patients. To ensure statistical independence of observations, only the first episode was analyzed for patients with multiple positive blood cultures. For those with repeatedly negative cultures, a single time point was randomly selected for inclusion. A flowchart outlining the patient selection process is provided in Fig.\u0026nbsp;1.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eOutcome\u003c/h3\u003e\n\u003cp\u003eThe outcome assessed was GN-BSI, defined as the growth of a gram negative bacteremia in at least one blood culture bottle[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e].\u003c/p\u003e\n\u003ch3\u003eDataset\u003c/h3\u003e\n\u003cp\u003eWe constructed datasets from the target medical centers comprising patient demographics, clinical and laboratory parameters. These variables were collected within a 24-hour window surrounding the time of blood culture collection. The dataset was structured to include the following variables: (1) patient demographics: age and sex; (2) past medical history: hypertension, diabetes mellitus, coronary heart disease, cerebrovascular disease, and solid cancer; (3) vital signs: temperature, heart rate; respiratory rate,and blood pressure; (4) antibiotic usage prior to pathogen detection; (5) traumatic operation: gastric tube insertion, urethral catheterization, endotracheal tube, tracheotomy, deep vein catheterization\u0026zwnj;, arterial catheterization, drainage tube, continuous renal replacement therapy, and recent surgical procedures\u0026zwnj;\u0026zwnj;; (6) laboratory parameters:blood cells,hemagglutination,liver function,renal function; electrolytes,inflammatory markers, serum glucose, blood gas analysis ,and blood culture.\u003c/p\u003e\n\u003ch3\u003eData preprocessing\u003c/h3\u003e\n\u003cp\u003eData cleaning and preprocessing are essential phases in the data analysis pipeline, designed to convert raw data into a structured dataset suitable for robust statistical analysis or machine learning modeling[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. To address missing data, we excluded variables where the proportion of missing values exceeded 15%[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. For the remaining missing data, we employed a multiple imputation approach using the \"MICE\" package in R to generate complete datasets for robust analysis [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e].\u003c/p\u003e\n\u003ch3\u003eFeature selection\u003c/h3\u003e\n\u003cp\u003eA pre-seeded random number generator (123) in R software was utilized to randomly divide the cohort into training and validation sets based on a ratio of 7:3.The training sets were used for modelling,while the validation sets for internal validation. We employed an L1-penalty least absolute shrinkage and selection operator (LASSO) regression approach to screen variables, augmented with 10-fold cross-validation[[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. LASSO regression is a method used to reduce the dimensionality of data by selecting features based on a penalty function. It effectively eliminate multicollinearity and avoid over-fitting of variables. Subsequently, we incorporate the selected variables in LASSO into a multivariate logistic regression analysis to identify the predictive variables for GN-BSI in ICU patients and construct predictive models.\u003c/p\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eMachine learning algorithms\u003c/h2\u003e\u003cp\u003eFollowing variable selection, we employed a suite of machine learning algorithms, including Logistic Regression, Decision Tree, Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), Support Vector Machine (SVM), and Artificial Neural Network (ANN) to predict the risk of GN-BSI in ICU patients. Throughout the model development phase, we use a grid search technique and 5-fold cross validation to to tune the hyperparameters and derive the optimal model for each algorithm.\u003c/p\u003e\u003cp\u003eThe predictive performance of each model was evaluated using a comprehensive set of metrics, including accuracy, sensitivity, specificity, F1 score, positive predictive value (PPV), negative predictive value (NPV), and the area under the receiver operating characteristic curve (AUC). The optimal model was further evaluated on the validation set using a calibration curve to assess the agreement between its predictions and the actual GN-BSI outcomes, as well as by decision curve analysis (DCA) to determine its net clinical benefit.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eSHAP interpretability analysis\u003c/h3\u003e\n\u003cp\u003eThe \"black-box\" nature of many machine learning models often obscures the influence of individual risk variables on their predictions. To enhance interpretability, we employed Shapley Additive Explanation (SHAP) values to precisely quantify the contribution and significance of each feature to the predictions generated by our optimal model[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e].The higher SHAP value indicates great impact of a feature on model output. We employed a SHAP bar and beeswarm plots to evaluate feature importance, followed by utilizing SHAP dependency plot to investigate the impact of features on outcome prediction. Finally, a SHAP force analysis was used to elucidate the contribution of features in individual patients.\u003c/p\u003e\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003eStatistical analysis\u003c/h2\u003e\u003cp\u003eContinuous variables are presented as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SEM (range) or median (interquartile range). Comparisons were conducted using either the Student\u0026rsquo;s t-test or the Wilcoxon rank-sum test. Categorical variables are expressed in terms of frequencies and percentages, with comparisons performed using the chi-square test or Fisher\u0026rsquo;s exact test, as appropriate. A two-sided P-value of less than 0.05 was deemed statistically significant. Statistical analyses were carried out utilizing SPSS version 27.0 (IBM Corp), R version 4.5.1 (The R Foundation for Statistical Computing), and Python version 3.10.4 (Python Software Foundation).\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003ePatient characteristics\u003c/h2\u003e\u003cp\u003eDuring the study period, 596 patients were admitted to ICU. After applying the exclusion criteria,which comprised patients under 18 years of age (n\u0026thinsp;=\u0026thinsp;4), those with an ICU stay of less than 48 hours (n\u0026thinsp;=\u0026thinsp;79), and those with blood cultures indicating non-gram-negative bacteria (n\u0026thinsp;=\u0026thinsp;108),a total of 405 patients were included in the study. From this final cohort, 94 patients were identified with GN-BSI, accounting for 23.21%. Variables with a missing data ratio exceeding 15% (namely, PH, PO2, PCO2,BE, and Lac) were excluded from further analysis. The distribution of missing data for all variables is detailed in Supplementary Fig.\u0026nbsp;1 and Supplementary Table\u0026nbsp;1. The included patients were divided into the training set (284 patients) and the validation set (121 patients).In the training and validation set, the median age was 61 (IQR:49, 72) and 61.5 (IQR:53, 72) years, and 195 (62.7%) and 60 (63.83%) patients were men,respectively. A total of 81 variables were collected for each patient. The detailed features between individuals with and without GN-BSI were summarized in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eBaseline characteristics of ICU patients with and without GN-BSI\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVariables\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eOverall(n\u0026thinsp;=\u0026thinsp;405)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNon-GN-BSI (n\u0026thinsp;=\u0026thinsp;311)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eGN-BSI (n\u0026thinsp;=\u0026thinsp;94)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cem\u003ep-\u003c/em\u003evalue\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\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\u003e255 (62.96)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e195 (62.70)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e60 (63.83)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.939\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHypertension (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e153 (37.78)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e115 (36.98)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e38 (40.43)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.629\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDiabetes mellitus (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e63 (15.56)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e45 (14.47)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e18 (19.15)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.35\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCoronary heart disease (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e39 ( 9.63)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e33 (10.61)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e6 ( 6.38)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.309\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCerebrovascular disease (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e40 ( 9.88)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e31 ( 9.97)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e9 ( 9.57)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSolid cancer (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e38 ( 9.38)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e26 ( 8.36)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e12 (12.77)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.279\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRecent surgical operation (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e242 (59.75)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e186 (59.81)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e56 (59.57)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003elactamase inhibitor(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e229 (56.54)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e179 (57.56)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e50 (53.19)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.529\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCarbapenem antibiotics (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e122 (30.12)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e64 (20.58)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e58 (61.70)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eColistin (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e11 ( 2.72)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e2 ( 0.64)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e9 ( 9.57)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTetracycline (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e10 ( 2.47)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e3 ( 0.96)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e7 ( 7.45)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.002*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGlycopeptides antibiotics(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e88 (21.73)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e47 (15.11)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e41 (43.62)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAntifungal drugs (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e42 (10.37)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e16 ( 5.14)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e26 (27.66)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eImmunosuppressive drugs (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e18 ( 4.44)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e7 ( 2.25)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e11 (11.70)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGastric tube insertion (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e212 (52.35)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e155 (49.84)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e57 (60.64)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.086\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eUrethral catheterization (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e203 (50.12)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e142 (45.66)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e61 (64.89)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.002*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEndotracheal intubation (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e310 (76.54)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e239 (76.85)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e71 (75.53)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.9\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTracheotomy(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e40 ( 9.88)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e26 ( 8.36)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e14 (14.89)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.096\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDeep vein catheterization (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e160 (39.51)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e104 (33.44)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e56 (59.57)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eArterial Catheterization (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e109 (26.91)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e72 (23.15)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e37 (39.36)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.003*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDrainage tube (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e206 (50.86)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e157 (50.48)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e49 (52.13)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.871\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCRRT (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e48 (11.85)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e19 ( 6.11)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e29 (30.85)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge (years)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e61.00 [50.00, 72.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e61.00 [49.00, 72.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e61.50 [53.00, 72.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.688\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTemperature (℃)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e37.10 [36.60, 37.80]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e37.00 [36.60, 37.80]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e37.30 [36.80, 38.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.115\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSBP ((mmHg)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e126.00[116.00,138.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e127.00 [117.00138.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e122.00[110.00,135.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.016*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDBP ((mmHg)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e70.00 [64.00, 78.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e70.00 [65.00, 78.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e70.00 [62.00, 76.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.309\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHeart rate (min\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e91.00 [80.00, 103.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e90.00 [80.00, 102.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e94.50 [82.00, 110.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.069\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRespiratory rate (min\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e18.00 [15.00, 20.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e18.00 [15.00, 20.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e18.00 [15.00, 20.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.754\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePT (s)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e15.50 [14.40, 17.10]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e15.30 [14.30, 16.75]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e16.70 [15.00, 19.10]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePT %(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e73.00 [61.00, 84.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e75.00 [63.00, 85.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e64.00 [50.00, 77.75]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eINR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.21 [1.11, 1.37]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1.19 [1.10, 1.35]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.33 [1.16, 1.60]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAPTT (s)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e39.30 [34.70, 45.30]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e38.20 [34.10, 43.70]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e41.70 [38.15, 48.32]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFIB (g/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e4.36 [2.74, 5.96]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e4.34 [2.70, 6.04]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e4.48 [2.94, 5.78]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.506\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTT (s)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e17.00 [15.90, 18.60]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e16.90 [15.90, 18.45]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e17.25 [15.80, 19.08]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.311\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eD-D ( \u0026micro;g/mL)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e2.95 [1.80, 9.13]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e2.85 [1.71, 8.84]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e3.62 [2.03, 9.96]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.176\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFDP (\u0026micro;g/mL)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e12.71 [6.60, 33.65]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e12.10 [6.30, 31.56]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e15.53 [8.00, 37.23]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.047*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWBC (10\u003csup\u003e9\u003c/sup\u003e/L )\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e11.11 [7.85, 15.66]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e11.11 [7.94, 15.09]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e11.16 [7.62, 16.96]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.615\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNeu% (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e87.90 [83.20, 91.20]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e87.00 [82.70, 90.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e89.70 [86.03, 93.65]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLym% (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e6.40 [3.90, 10.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e6.70 [4.10, 10.40]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e5.30 [3.02, 8.20]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMon% (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e4.80 [3.10, 6.60]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e5.10 [3.50, 6.80]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e3.55 [2.02, 5.25]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEos% (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.10 [0.00, 0.60]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.10 [0.00, 0.70]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.10 [0.00, 0.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.986\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBas% (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.20 [0.10, 0.30]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.20 [0.10, 0.30]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.20 [0.10, 0.30]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.177\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNeu (10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e9.79 [6.81, 13.75]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e9.69 [6.83, 13.07]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e10.13 [6.38, 15.59]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.337\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLym (10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.69 [0.42, 1.09]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.73 [0.44, 1.17]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.57 [0.33, 0.79]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMon (10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.51 [0.29, 0.83]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.55 [0.33, 0.87]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.41 [0.20, 0.66]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEos (10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.01 [0.00, 0.07]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.01 [0.00, 0.07]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.01 [0.00, 0.06]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.666\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBas (10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.02 [0.01, 0.03]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.02 [0.01, 0.03]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.02 [0.01, 0.04]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.178\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRBC (10\u003csup\u003e12\u003c/sup\u003e/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e3.21 [2.65, 3.88]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e3.20 [2.66, 3.94]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e3.25 [2.62, 3.75]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.517\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHb (g/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e98.00 [82.00, 119.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e98.00 [82.00, 119.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e98.50 [81.25, 114.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.405\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHCT (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e30.10 [25.00, 35.90]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e30.30 [25.00, 36.55]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e29.40 [24.92, 34.25]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.336\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMCV (fL)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e93.50 [89.80, 97.40]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e93.50 [89.80, 97.30]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e93.50 [89.32, 97.68]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.765\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMCH (pg )\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e30.50 [29.40, 31.80]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e30.50 [29.30, 31.80]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e30.50 [29.63, 31.60]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.991\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMCHC ( g/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e327.00[316.00,337.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e328.00[316.00,337.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e326.00[316.50,338.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.705\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRDW -CV (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e14.00 [13.20, 15.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e13.90 [13.20, 15.35]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e14.60 [13.30, 15.88]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.01*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRDW-SD (fL)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e47.80 [43.80, 52.10]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e47.50 [43.50, 51.85]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e48.80 [45.02, 53.08]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.115\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePLT (10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e140.00 [81.00, 207.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e156.00[94.00,213.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e96.50 [57.25, 159.75]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePCT (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.16 [0.10, 0.23]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.18 [0.11, 0.24]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.12 [0.07, 0.19]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePDW (fL)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e14.20 [12.10, 16.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e14.00 [11.95, 16.10]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e15.10 [12.80, 17.78]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.009*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMPV (fL)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e11.50 [10.70, 12.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e11.50 [10.60, 12.40]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e11.80 [10.90, 12.80]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.025*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTP (g/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e56.87 (8.62)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e57.34 (8.70)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e55.34 (8.20)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.049*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALB (g/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e34.34 (5.74)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e34.74 (5.77)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e33.02 (5.49)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.011*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGLB (g/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e22.49 (5.76)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e22.54 (5.83)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e22.32 (5.57)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.754\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eA/G\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.53 [1.31, 1.82]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1.56 [1.31, 1.83]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.47 [1.24, 1.77]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.24\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTBIL (\u0026micro;mol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e17.20 [11.04, 30.22]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e15.85 [10.27, 27.29]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e24.65 [13.48, 63.41]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDBIL (\u0026micro;mol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e5.16 [2.86, 12.46]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e4.30 [2.56, 10.20]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e9.71 [4.90, 37.19]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIBIL (\u0026micro;mol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e11.28 [7.50, 17.95]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e10.62 [7.32, 16.30]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e14.33 [7.88, 30.06]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.003*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALT (U/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e32.50 [17.50, 100.70]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e31.00 [16.55, 96.90]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e38.90 [19.60, 146.35]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.233\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAST (U/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e46.00 [29.20, 102.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e47.00 [29.45, 102.20]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e44.90 [28.95, 106.03]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.97\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALP (U/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e83.80 [63.60, 110.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e80.20 [60.80, 104.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e95.90 [67.45, 128.30]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.003*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGGT (U/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e42.40 [21.20, 79.40]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e39.20 [20.05, 76.05]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e47.60 [29.30, 106.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.054*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBUN (mmol/L )\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e9.50 [6.38, 14.95]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e9.03 [6.14, 14.25]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e12.36 [7.89, 17.67]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCRE (\u0026micro;mol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e78.60 [57.40, 131.60]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e75.10 [55.25, 117.35]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e90.90 [67.73, 197.33]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eUA (\u0026micro;mol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e270.30[165.30,399.10]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e272.80[169.30,389.15]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e251.20[155.43,424.72]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.773\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eeGFR(ml/(min\u0026middot;1.73m\u003csup\u003e2\u003c/sup\u003e))\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e83.00 [46.00, 104.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e86.00 [55.50, 105.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e66.00 [26.25, 96.75]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.002*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGLU (mmol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e8.50 [6.46, 11.30]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e8.44 [6.58, 11.12]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e9.04 [6.13, 11.74]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.868\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eK (mmol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e4.06 [3.70, 4.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e4.05 [3.69, 4.48]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e4.16 [3.72, 4.56]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.627\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNa (mmol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e141.10[137.60,146.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e141.30[137.95,146.50]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e140.75[136.62,146.38]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.428\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCl (mmol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e105.10[101.40,110.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e105.20[101.80,110.00]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e104.25[100.70,109.95]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.276\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHCO3 (mmol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e24.30 [21.70, 27.10]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e24.50 [22.20, 27.20]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e22.90 [19.83, 26.58]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.003*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eProcalcitonin (ng/mL)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.05 [0.33, 5.58]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.64 [0.24, 1.90]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e14.13 [4.10, 33.67]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCRP ( mg/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e57.90[15.20,139.47]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e41.32 [10.00, 99.51]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e149.87 [84.61, 178.24]\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"5\"\u003eAbsolute numbers and percentages are used for categorical variables and mean and standard deviation are used for continuous variables\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd colspan=\"5\"\u003e*shows the significant differences between the GN-BSI and Non-GN-BSI\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003eVariables of importance\u003c/h2\u003e\u003cp\u003eIn the training set, the optimal lambda value was determined to be 0.059 using LASSO regression (Fig.\u0026nbsp;2A and B), with seven variables identified as potential predictors from the baseline characteristics. These variables are listed in detail in Supplementary Table\u0026nbsp;2.Subsequently, the seven variables selected by LASSO regression were incorporated into a multivariate logistic regression model. Of these, four variables shown in the Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e were retained as optimal predictive features: deep vein catheterization, continuous renal replacement therapy (CRRT), procalcitonin, and C-reactive protein (CRP).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003eMultimodel integrated analysis for classification\u003c/h2\u003e\u003cp\u003eAll seven machine learning models demonstrated excellent predictive performance in the training set(Supplementary Table\u0026nbsp;3). For the hyperparameters of each model,please refer to Supplementary Table\u0026nbsp;4 in the supplementary material. In the validation set,the ROC curves and confusion matrices of each model are shown in Fig.\u0026nbsp;3A and Fig.\u0026nbsp;4, and all models achieved accuracy and AUC values of 0.85 or above. From the evaluation metrics in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e,the XGBoost model demonstrated comprehensive superiority, achieving the highest F1-score,a metric that balances precision and sensitivity,and outperforming all other models in sensitivity, specificity, Kappa value, Youden's J statistic, and positive and negative predictive values. The XGBoost model demonstrated excellent predictive performance, with an AUC of 0.898 (95% CI: 0.807\u0026ndash;0.959), an accuracy of 88.43%, a precision of 85.00%, and a sensitivity of 60.71%. Additionally, it achieved a specificity of 96.77%, an F1-score of 70.83%, a PPV of 85.00%, and an NPV of 89.10%.Considering the F1 score and the other evaluation metrics, the XGBoost model emerged as the best model. The Brier score of the calibration curve for the XGBoost model is 0.101, indicating that they have relatively excellent prediction performance (Fig.\u0026nbsp;3B). According to the DCA curve (Fig.\u0026nbsp;3C), the XGBoost model showed a large net benefit,indicating that the established model has robust clinical validity.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003eInterpretability and application of the model\u003c/h2\u003e\u003cp\u003eThe feature\u0026rsquo;s contribution to the prediction outcomes was quantified using SHAP, which provides more insights into how the XGBoost model predicted outcomes.The SHAP feature importance ranking depict the standard bar chart of the mean absolute SHAP value for each feature in descending order(Fig.\u0026nbsp;5A). The SHAP beeswarm plot provides insights into the predictions of the XGBoost model. The diagram illustrates the degree of influence each feature has on model output(Fig.\u0026nbsp;5B). The color gradient (feature value)from red to blue represents the magnitude of the feature value, with red denoting a high feature value and blue indicating a low feature value. A positive Shapley value for each feature indicates an increased risk of GN-BSI while a negative value suggests decreased risk. Figure\u0026nbsp;6 is SHAP dependency plot,which analyze the impact of features at factor level on the risk of the predictive model. It demonstrated that higher PCT and CRP values, performed deep vein catheterization and CRRT contributed to an elevated risk of GN-BSI in ICU patients.Additionally, the force plots provided personalized feature attributions using two representative examples and illustrated how to use SHAP to interpret the predictions of individual model, as shown in Fig.\u0026nbsp;7A (a ICU patient without GN-BSI) and B (a ICU patient with GN-BSI).\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;1.A flowchart describing the number of patients included in the analysis after exclusion criteria. The ICU patients included were divided into those with and without GN-BSI\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;2.Presentation of the results of the LASSO regression analysis. (A) LASSO Regression Model Factor Selection: Left line represents the optimal lambda value (lambda\u0026sdot;min), while the right line marks the lambda value within one standard error of the optimal (lambda.1se); (B) LASSO regression model\u003c/p\u003e\u003cp\u003escreening variable trajectories.\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;3.The performance and comparison of seven different predictive models, (A) ROC curves and AUC values of the validation set (B) Calibration curves of the XGBoost in the validation set (C) Decision curves analysis of the XGBoost in the validation set.\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;4.Confusion Matrices of ML models. (A) Logistic Regression\u0026rsquo;s Confusion Matrices. (B) Decision Tree\u0026rsquo;s Confusion Matrices. (C) Random Forest\u0026rsquo;s Confusion Matrices. (D) XGBoost\u0026rsquo;s Confusion Matrices. (E) LightGBM\u0026rsquo;s Confusion Matrices. (F) SVM\u0026rsquo;s Confusion Matrices. (G) ANN\u0026rsquo;s Confusion Matrices.\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\u003eMultivariate logistic regression analysis of GN-BSI in ICU patients\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"7\"\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\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVariables\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eβ\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSE\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eWald\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eOR\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003e95%CI\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cem\u003eP-\u003c/em\u003evalue\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCarbapenem antibiotics\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.507\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.454\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.117\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e1.66\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.682\u0026ndash;4.037\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.264\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDeep vein catheterization\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.903\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.428\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e2.108\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e2.466\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e1.065\u0026ndash;5.708\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.035*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCRRT\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.336\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.579\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e2.309\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e3.806\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e1.224\u0026ndash;11.834\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.021*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAPTT\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.03\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.016\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.84\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e1.03\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.998\u0026ndash;1.064\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.066\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDBIL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.006\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.004\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.385\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e1.006\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.998\u0026ndash;1.014\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.166\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eProcalcitonin\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.074\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.018\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e4.163\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e1.077\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e1.04\u0026ndash;1.115\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\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\u003eCRP\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.009\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.003\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e2.743\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e1.009\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e1.003\u0026ndash;1.016\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.006*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"7\"\u003e*shows the significant differences between the GN-BSI and Non-GN-BSI\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eEvaluation metrics results of seven models in the validation set\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=\"left\" 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=\"char\" char=\".\" 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\u003eAUC(95%CI)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eAccuracy\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSensitivity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eSpecificity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eF1 Score\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e\u003cp\u003eKappa\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c8\"\u003e\u003cp\u003eYouden's J\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c9\"\u003e\u003cp\u003ePPV\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c10\"\u003e\u003cp\u003eNPV\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLogistic\u003c/p\u003e\u003cp\u003eregression\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.889\u003c/p\u003e\u003cp\u003e(0.807\u0026ndash;0.955)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.851\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.957\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.609\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.522\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.457\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.778\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e\u003cp\u003e0.864\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDecision Tree\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.877\u003c/p\u003e\u003cp\u003e(0.790\u0026ndash;0.946)\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\u003e0.643\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.946\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.706\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.628\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.589\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.783\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e\u003cp\u003e0.898\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRandom Forest\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.874\u003c/p\u003e\u003cp\u003e(0.790\u0026ndash;0.941)\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\u003e0.536\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.935\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.612\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.516\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.471\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.714\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e\u003cp\u003e0.87\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=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.898\u003c/p\u003e\u003cp\u003e(0.807\u0026ndash;0.959)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.884\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.607\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.968\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.708\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.639\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.575\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.85\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e\u003cp\u003e0.891\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=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.872\u003c/p\u003e\u003cp\u003e(0.782\u0026ndash;0.941)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.826\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.571\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.903\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.604\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.493\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.475\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.64\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e\u003cp\u003e0.875\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=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.910\u003c/p\u003e\u003cp\u003e(0.826\u0026ndash;0.971)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.868\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.978\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.636\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.563\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.478\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.875\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e\u003cp\u003e0.867\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eANN\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.911\u003c/p\u003e\u003cp\u003e(0.834\u0026ndash;0.968)\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\u003e0.643\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.946\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.706\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.628\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.589\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.783\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e\u003cp\u003e0.898\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\u003eFigure\u0026nbsp;5.Feature importance analysis by SHAP method (A) The SHAP feature importance ranking plot of features of the XGBoost model. (B) SHAP beeswarm plot of the XGBoost model.\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;6.SHAP dependency plot of features in the XGBoost model. The Y-axis represents SHAP values, while the X-axis represents actual clinical parameters.For binary variables such as Deep vein catheterization and CRRT, \u0026ldquo;0\u0026rdquo; indicates the absence of the condition, while \u0026ldquo;1\u0026rdquo; indicates its presence.. Significantly, when a feature\u0026rsquo;s SHAP value is greater than 0, it suggests an increased risk of GN-BSI, whereas a negative SHAP value suggests a reduced risk.\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;7.SHAP force plot of features in the XGBoost model. The force plots provide personalized feature attributions using two representative examples. A: a ICU patient without GN-BSI; B: a ICU patient with GN-BSI.\u003c/p\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eConstructing a predictive model for Gram-negative bacterial bloodstream infection (GNB-BSI) in the ICU is critical for the early identification of pathogens, improving patient survival rates, and combating antimicrobial resistance. To this end, this study developed and compared several prediction models using a range of machine learning algorithms. Among them, the eXtreme Gradient Boosting (XGBoost) model demonstrated the most favorable predictive value, confirming its utility as an effective tool for early clinical intervention and for delaying the development of bacterial resistance in this high-risk patient population.\u003c/p\u003e\u003cp\u003eAs far as we know, Feature selection is a crucial process in the development of predictive models. In this study, Lasso regression was first employed to identify seven key predictors. Following this, multivariate logistic regression was subsequently applied to refine the selection, ultimately identifying four final variables, namely deep vein catheterization, CRRT, Procalcitonin, and CRP. The relationship between these features and BSI has been extensively examined in prior research. Regarding deep vein catheterization, it is a significant risk factor for hospital-acquired bloodstream infections[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. The catheter, as a foreign body, serves as a conduit for skin flora (both resident and transient) to enter the bloodstream. Simultaneously, a biofilm rapidly forms on its surface, providing a niche for bacterial colonization. This biofilm confers antibiotic resistance and is a common cause of recurrent bacteremia. Among high-risk populations such as intensive care and hemodialysis patients, the incidence of catheter-related bloodstream infections (CRBSI) ranges from approximately 0.5 to 5.0 per 1000 catheter-days [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e], with Gram-negative bacteria, including \u003cem\u003eEscherichia coli\u003c/em\u003e and \u003cem\u003eKlebsiella\u003c/em\u003e species, accounting for approximately 30% of the causative pathogens[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e].A biological link has been established between CRRT in ICU and GNB-BSI, with research indicating that patients requiring CRRT are typically critically ill, immunocompromised individuals in the ICU, which predisposes them to bacterial infections[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. Their heightened susceptibility to bloodstream infections is largely attributable to the prolonged use of central venous catheters, resulting in a significantly greater risk compared to non-CRRT patients or general nursing populations. While Gram-positive bacteria remain the primary pathogens, the incidence of Gram-negative infections has risen notably in recent years[\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. Gatti M et al[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. also pointed out that end-stage renal disease and central catheterization are independent risk factors for GN-BSI. Previous studies have shown that PCT and CRP are potential biomarkers for BSI[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. In healthy individuals,PCT levels are typically undetectable or very low (\u0026lt;\u0026thinsp;0.05 ng/mL). However, during severe systemic bacterial infections like sepsis or bacteremia, endotoxins and inflammatory cytokines (including TNF-α, IL-6) trigger a substantial release of PCT from iver macrophages and systemic tissue cells throughout the body, resulting in a marked elevation of serum concentrations[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. Furthermore, PCT levels are typically more elevated in Gram-negative bloodstream infections[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e], suggesting their potential utility in predicting the causative pathogen type. CRP is an acute-phase reactant whose levels rise rapidly in response to a wide range of conditions, including inflammation, infection (bacterial, viral, or fungal), tissue damage, and malignancy. Consequently, it lacks the specificity required for diagnosing bloodstream infections, and its diagnostic performance in this context is inferior to that of PCT[\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. Therefore, the combined application of PCT and CRP plays a synergistic role in significantly enhancing the diagnostic accuracy for bloodstream infections and facilitating earlier clinical intervention[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eOur results indicate that the XGBoost model demonstrated superior overall performance compared to the other six machine learning algorithms across key metrics, including accuracy, sensitivity, specificity, AUROC values and F1-score in the validation set, as did XGBoost in the training set. According to the DCA of the training and validation sets, the intervention measures guided by the predictive model produced outstanding results, except for a small range of low preferences. Furthermore, the model calibration was evaluated using a calibration curve, and the Brier score of XGBoost in the validation set was 0.102 (95% CI 0.070\u0026ndash;0.137), demonstrating strong calibration and prediction accuracy. Based on its superior discriminative ability, well-calibrated probabilistic outputs, and significant clinical net benefit demonstrated through comprehensive evaluation, the XGBoost algorithm was selected as the optimal predictive tool for Gram-negative bacteremia in ICU patients. Unlike previous models primarily designed for predicting generalized bloodstream infections, this study establishes a specialized framework for the early detection of GNB-BSI[\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. Given that GNB-BSI is frequently associated with rapid clinical deterioration, a heightened risk of septic shock, and significant antimicrobial resistance, the capacity for early and precise pathogen orientation offers critical decision support. This enables clinicians to initiate timely and targeted anti-Gram-negative therapy, thereby enhancing the model's practical utility in managing high-risk ICU patients.Moreover,this study comprehensively evaluated 81 variables encompassing demographic characteristics, relatively comprehensive set of clinical and laboratory parameters, ensuring extensive coverage of potential predictors for GNB-BSI in ICU patients.Compared to prior models requiring more variables,such as Hu et al. (5 variables) and Zhang et al. (32 variables),our model achieves a superior predictive power (XGBoost AUC: 0.898) using only 4 routine ICU indicators[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eOur model also has its limitations. Firstly, the generalizability of this study may be limited by its relatively small sample size, which carries a potential for selection bias. Secondly, although we have made every effort to collect potential predictive factors for GNB-BSI, some important risk factors, such as a large number of missing blood gas analysis indicators, must be excluded as related variables. These may lead us to overlook some features. Finally, the model demonstrated robust performance during internal validation; however, its generalizability requires confirmation through external validation. For clinical application, we recommend that future studies adopt a consistent analytical framework aligned with our research objectives. The predicted outcomes should be interpreted within the clinical context to ensure they contribute meaningfully to improving patient prognosis.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003ctable border=\"0\" cellpadding=\"0\" cellspacing=\"0\" width=\"396\" class=\"fr-table-selection-hover\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\" style=\"width: 66px;\"\u003eBSI\u003c/td\u003e\n \u003ctd class=\"xl65\" style=\"width: 330px;\"\u003eBloodstream Infection\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003eGN-BSI\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eGram-negative Bloodstream Infection\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003eICU\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eIntensive Care Unit\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003eML\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eMachine learning\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003eAUC\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eArea under the receiver operating characteristic curve\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003eSHAP\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eShapley Additive Explanation\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003eLASSO\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eL1-penalty least absolute shrinkage and selection operator\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003eRF\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eRandom Forest\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003eXGBoost\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eeXtreme Gradient Boosting\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003eLightGBM\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eLight Gradient Boosting Machine\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003eSVM\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eSupport Vector Machine\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003eANN\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eArtificial Neural Network\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003ePPV\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003ePositive predictive value\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003eNPV\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eNegative predictive value\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003eDCA\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eDecision curve analysis\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003eCRRT\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eContinuous renal replacement therapy\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eSBP\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eSystolic blood pressure\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eDBP\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eDiasto blood licpressure\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003ePT\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eProthrombin Time\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003ePT%\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eProthrombin Activity\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eINR\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eInternational Normalized Ratio\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eAPTT\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eActivated Partial Thromboplastin Time\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eFIB\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eFibrinogen\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eTT\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eThrombin Time\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eD-D\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eD-dimer\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eFDP\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eFibrinogen Degradation Products\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"20\" class=\"xl66\"\u003eWBC\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eWhite Blood Cell Count\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eNeu%\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eNeutrophils Percentage\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eLym%\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eLymphocyte Percentage\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eMon%\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eMonocytes Percentage\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eEos%\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eEosinophils Percentage\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eBas%\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eBasophils Percentage\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"20\" class=\"xl66\"\u003eNeu\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eNeutrophil Count\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eLym\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eLymphocyte Count\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eMon\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eMonocyte Count\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"20\" class=\"xl66\"\u003eEos\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eDirect Eosinophil Count\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eBas\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eDirect Basophil Count\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"20\" class=\"xl66\"\u003eRBC\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eRed Blood Cell Count\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eHb\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eHemoglobin\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eHCT\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eHematocrit\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eMCV\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eMean Corpuscular Volume\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eMCH\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eMean Corpuscular Hemoglobin\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eMCHC\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eMean Corpuscular Hemoglobin Concentration\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eRDW-CV\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eRed Blood Cell Distribution Width - Coefficient of Variation\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eRDW-SD\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eRed Blood Cell Distribution Width - Standard Deviation\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"20\" class=\"xl66\"\u003ePLT\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003ePlatelet Count\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003ePCT\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003ePlateletcrit\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003ePDW\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003ePlatelet Distribution Width\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eMPV\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eMean Platelet Volume\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eTP\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eTotal Protein\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eALB\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eAlbumin\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eGLB\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eGlobulin\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eA/G\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eAlb/Glb Ratio\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eTBIL\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eTotal Bilirubin\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eDBIL\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eDirect Bilirubin\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eIBIL\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eIndirect Bilirubin\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eALT\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eAlanine Aminotransferase\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eAST\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eAspartate Aminotransferase\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eALP\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eAlkaline Phosphatase\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eGGT\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eGamma-glutamyl Transferase\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eBUN\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eBlood Urea Nitrogen\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eCRE\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eCreatinine\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eUA\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eUric Acid\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"20\" class=\"xl66\"\u003eeGFR\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eEstimated Glomerular Filtration Rate\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eGLU\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eGlucose\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eK\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003ePotassium\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eNa\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eSodium\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eCl\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eChlorine\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eHCO3\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eBicarbonate\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl66\"\u003eCRP\u0026nbsp;\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003eC-Reactive Protein\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003ePH\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003ePotential of Hydrogen\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003ePO2\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003ePartial Pressure of Oxygen\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003ePCO2\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003ePartial Pressure of Carbon Dioxide\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl65\"\u003eBE\u003c/td\u003e\n \u003ctd class=\"xl65\"\u003ePartial Pressure of Carbon Dioxide\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"19\" class=\"xl67\"\u003eLac\u003c/td\u003e\n \u003ctd class=\"xl67\"\u003eLactic Acid\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"},{"header":"Declarations","content":"\u003ch2\u003eEthics approval and consent to participate\u003c/h2\u003e\n\u003cp\u003eThe study protocol received approval from the Clinical Medical Research Ethics Committee of the First Affiliated Hospital of Anhui Medical University (PJ2025-11-09) and was conducted in accordance with the principles of the Declaration of Helsinki. Due to the retrospective nature of the study, the requirement for written informed consent was waived. All patient data were de-identified prior to analysis to ensure confidentiality.\u003c/p\u003e\n\u003ch2\u003eConsent for the publication\u003c/h2\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003ch2\u003eCompeting interests\u003c/h2\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003ch2\u003eAuthors\u0026apos; information\u003c/h2\u003e\n\u003cp\u003e\u003csup\u003e1\u003c/sup\u003eDepartment of Clinical Laboratory, The First Affiliated Hospital of Anhui Medical University, Hefei, Anhui Province, China\u003c/p\u003e\n\u003cp\u003e\u003csup\u003e2\u003c/sup\u003eDepartment of Reproductive Medicine, The First Affiliated Hospital of Anhui Medical University, Hefei, Anhui Province, China\u003c/p\u003e\n\u003ch2\u003eFunding\u003c/h2\u003e\n\u003cp\u003eThis work was supported by the First Affiliated Hospital of Anhui Medical University Horizontal Research Project(CXPJJH21002-2021SS)\u003c/p\u003e\n\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\n\u003cp\u003eResearch design and conception were performed by Ya-Ling Zhou, Zhong-Le Cheng, and Zhong-Xin Wang. Data collection and statistical analysis were carried out by Ya-Ling Zhou，Hong-ting Da and Ting-Ting Wang. Ya-Ling Zhou developed the prediction system based on the proposed model. The initial draft of the manuscript was written by Ya-Ling Zhou, and Zhong-Le Cheng critically revised it for important intellectual content. All authors have reviewed and approved the final version of the manuscript.\u003c/p\u003e\n\u003ch2\u003eAcknowledgements\u003c/h2\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003ch2\u003eData Availability\u003c/h2\u003e\n\u003cp\u003eThe datasets used and analyzed in this study are available from the corresponding author upon reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eZheng C, Chen Q, Pan S, Li Y, Zhong L, Zhang X, Cui W, Lin R, Zhang G, Zhang S. Staphylococcus aureus bloodstream infection in a Chinese tertiary-care hospital: A single-center retrospective study. Chin Med J (Engl). 2023;136(12):1503\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKern WV, Rieg S. Burden of bacterial bloodstream infection-a brief update on epidemiology and significance of multidrug-resistant pathogens. Clin Microbiol Infect. 2020;26(2):151\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePradubkham T, Suwanpimolkul G, Gross AE, Nakaranurack C. Intravenous to oral transition of antibiotics for gram-negative bloodstream infection at a University hospital in Thailand: Clinical outcomes and predictors of treatment failure. PLoS ONE. 2022;17(9):e0273369.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAl-Hasan MN. Gram-Negative Bloodstream Infection: Implications of Antimicrobial Resistance on Clinical Outcomes and Therapy. Antibiot (Basel). 2020;9(12):922.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eScaglione G, Perego M, Colaneri M, Genovese C, Brivio F, Covizzi A, Viaggi B, Bandera A, Gori A, Finazzi S, Palomba E. Understanding the burden of antibiotic resistance: a decade of carbapenem-resistant Gram-negative bacterial infections in Italian intensive care units. Front Microbiol. 2024;15:1405390.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKundar R, Gokarn K, CRISPR-Cas System. A Tool to Eliminate Drug-Resistant Gram-Negative Bacteria. Pharmaceuticals (Basel). 2022;15(12):1498.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGu S, Xiong J, Peng S, Hu L, Zhu H, Xiao Y, Luo H, Hang Y, Chen Y, Fang X, Cao X, Fang Y, Li F, Zhu J, Zhong Q. Assessment of Effective Antimicrobial Regimens and Mortality-Related Risk Factors for Bloodstream Infections Caused by Carbapenem-Resistant \u003cem\u003eAcinetobacter baumannii\u003c/em\u003e. Infect Drug Resist. 2023;16:2589\u0026ndash;600.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWarren BG, Yarrington ME, Polage CR, Anderson DJ, Moehring RW. Evaluation of hospital blood culture utilization rates to identify opportunities for diagnostic stewardship. Infect Control Hosp Epidemiol. 2023;44(2):200\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTimsit JF, Rupp\u0026eacute; E, Barbier F, Tabah A, Bassetti M. Bloodstream infections in critically ill patients: an expert statement. Intensive Care Med. 2020;46(2):266\u0026ndash;84.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLynch CJ, Liston C. New machine-learning technologies for computer-aided diagnosis. Nat Med. 2018;24(9):1304\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRichens JG, Lee CM, Johri S. Author Correction: Improving the accuracy of medical diagnosis with causal machine learning. Nat Commun. 2021;12(1):2108.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang F, Wang H, Liu L, Su T, Ji B. Machine learning model for the prediction of gram-positive and gram-negative bacterial bloodstream infection based on routine laboratory parameters. BMC Infect Dis. 2023;23(1):675.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHu X, Zhi S, Li Y, Cheng Y, Fan H, Li H, Meng Z, Xie J, Tang S, Li W. Development and application of an early prediction model for risk of bloodstream infection based on real-world study. BMC Med Inf Decis Mak. 2025;25(1):186.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRoimi M, Neuberger A, Shrot A, Paul M, Geffen Y, Bar-Lavie Y. Early diagnosis of bloodstream infections in the intensive care unit using machine-learning algorithms. Intensive Care Med. 2020;46(3):454\u0026ndash;62.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChicco D, Oneto L, Tavazzi E. Eleven quick tips for data cleaning and feature engineering. PLoS Comput Biol. 2022;18(12):e1010718.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMaletzky A, B\u0026ouml;ck C, Tschoellitsch T, Roland T, Ludwig H, Thumfart S, Giretzlehner M, Hochreiter S, Meier J. Lifting Hospital Electronic Health Record Data Treasures: Challenges and Opportunities. JMIR Med Inf. 2022;10(10):e38557.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMa C, Sun G, Yang X, Yang S. A clinically applicable prediction model for the risk of in-hospital mortality in solid cancer patients admitted to intensive care units with sepsis. J Cancer Res Clin Oncol. 2023;149(10):7175\u0026ndash;85.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFaquih T, van Smeden M, Luo J, le Cessie S, Kastenm\u0026uuml;ller G, Krumsiek J, Noordam R, van Heemst D, Rosendaal FR, van Hylckama Vlieg A, van Willems K, Mook-Kanamori DO. A Workflow for Missing Values Imputation of Untargeted Metabolomics Data. Metabolites. 2020;10(12):486.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSun K, Huang SH, Wong DSH, Jang SS. Design and application of a variable selection method for multilayer perceptron neural network with LASSO. IEEE Trans Neural Netw Learn Syst. 2017;28:1386\u0026ndash;96.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFeng S, Huang S, Zhou Z, Zhang B, Huang C, Chen T, Zhou C, Wu S, Zhu J, Chen J, Xue J, Zhan X, Liu C. Development and Validation of a Machine Learning-Based Online Prognostic Model for Cervical Spondylosis Patients After Anterior Cervical Discectomy and Fusion: A Multicenter Study. JOR Spine. 2025;8(3):e70090.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGuo Z, Wang T, Chi S, Huang L. Energy consumption analysis and prediction in exercise training based on accelerometer sensors and deep learning. Sci Rep. 2025;15(1):19423.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAbubakar U. Point-prevalence survey of hospital acquired infections in three acute care hospitals in Northern Nigeria. Antimicrob Resist Infect Control. 2020;9(1):63.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePitiriga V, Bakalis J, Theodoridou K, Kanellopoulos P, Saroglou G, Tsakris A. Lower risk of bloodstream infections for peripherally inserted central catheters compared to central venous catheters in critically ill patients. Antimicrob Resist Infect Control. 2022;11(1):137.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNeoh KG, Li M, Kang ET, et al. Surface modification strategies for combating catheter-related complications: recent advances and challenges[J]. J Mater Chem B. 2017;5(11):2045\u0026ndash;67.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBrunelli SM, Turenne W, Sibbel S, et al. Clinical and economic burden of bloodstream infections in critical care patients with central venous catheters[J]. J Crit Care. 2016;35:69\u0026ndash;74.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCetin S, Dokmetas I, Hamidi AA, Bayraktar B, Gunduz A, Sevgi DY. Comparison of Risk Factors and Outcomes in Carbapenem-Resistant and Carbapenem-Susceptible Gram-Negative Bacteremia. Sisli Etfal Hastan Tip Bul. 2021;55(3):398\u0026ndash;404.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGatti M, Bonazzetti C, Tazza B, Pascale R, Miani B, Malosso M, Beci G, Marzolla D, Rinaldi M, Viale P, Giannella M. Impact on clinical outcome of follow-up blood cultures and risk factors for persistent bacteraemia in patients with gram-negative bloodstream infections: a systematic review with meta-analysis. Clin Microbiol Infect. 2023;29(9):1150\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eStoller J, Halpin L, Weis M, Aplin B, Qu W, Georgescu C, et al. Epidemiology of severe sepsis: 2008\u0026ndash;2012. J Crit Care. 2016;31(1):58\u0026ndash;62.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eShim YE, Shin S, Ko Y, Kim DH, Lim SJ, Jung JH, et al. Serum procalcitonin as a biomarker for differentiating between infectious and non-infectious Fever after pancreas transplantation. Clin Transpl. 2021;35(4):e14224.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang X, Zeng J, Yu X, Wang Z, Wang D, Zhou Q, Bai T, Xu Y, PCT. IL-6, and IL-10 facilitate early diagnosis and pathogen classifications in bloodstream infection. Ann Clin Microbiol Antimicrob. 2023;22(1):103.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eShokripour M, Omidifar N, Salami K, Moghadami M, Samizadeh B. Diagnostic Accuracy of Immunologic Biomarkers for Accurate Diagnosis of Bloodstream Infection in Patients with Malignancy: Procalcitonin in Comparison with C-Reactive Protein. Can J Infect Dis Med Microbiol. 2020;2020:8362109.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiang P, Yu F. Value of CRP, PCT, and NLR in Prediction of Severity and Prognosis of Patients With Bloodstream Infections and Sepsis. Front Surg. 2022;9:857218.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHu X, Zhi S, Li Y, Cheng Y, Fan H, Li H, Meng Z, Xie J, Tang S, Li W. Development and application of an early prediction model for risk of bloodstream infection based on real-world study. BMC Med Inf Decis Mak. 2025;25(1):186.\u003c/span\u003e\u003c/li\u003e\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":"bmc-medical-informatics-and-decision-making","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"midm","sideBox":"Learn more about [BMC Medical Informatics and Decision Making](http://bmcmedinformdecismak.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/midm/default.aspx","title":"BMC Medical Informatics and Decision Making","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Gram-negative bloodstream infection, Intensive care unit, Machine learning, Prediction model","lastPublishedDoi":"10.21203/rs.3.rs-8240564/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8240564/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackgroud\u003c/h2\u003e\u003cp\u003eGram-negative bloodstream infection (GN-BSI) can induce fatal septic shock, and the increasingly severe problem of antimicrobial resistance results in high clinical mortality particularly in intensive care unit (ICU) patients. The early identification of pathogens and timely antibiotic therapy are critical for patient outcomes. However, conventional diagnostic methods like blood culture are time-consuming and can delay treatment. Furthermore, the the implementation of molecular detection techniques in routine laboratories is often hindered by high costs and technical complexity.Machine learning (ML) offers a promising alternative for early prediction of GN-BSI. This study aims to develop an early prediction model for GN-BSI by integrating clinical and laboratory parameters from ICU patients using machine learning algorithms, thereby assisting in the early diagnosis and treatment of GN-BSI.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eThis retrospective study utilized data from ICU patients admitted to the West District of the First Affiliated Hospital of Anhui Medical University between January and July 2025. Following data preprocessing and multiple imputation of missing values, the dataset was randomly divided into training and validation sets in a 7:3 ratio. Feature selection was performed using Lasso regression and multivariate logistic regression. Seven ML models were developed and evaluated based on metrics including the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, specificity,F1-score, positive predictive value (PPV), and negative predictive value (NPV). Model interpretability was further assessed using Shapley Additive Explanation (SHAP) analysis.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eThis study ultimately included 405 ICU patients. Following further feature selection, four variables were identified, including deep vein catheterization, continuous renal replacement therapy (CRRT), procalcitonin, and c-reactive protein (CRP). Early prediction models for GN-BSI in ICU patients were constructed using seven machine learning algorithms. Among them, the XGBoost model demonstrated the best performance, with the AUC value of 0.898, accuracy of 88.43%, F1 score of 0.783,PPV of 85.00%, and NPV of 89.10%. SHAP bar and beeswarm plots illustrate the contribution of the four variables to the outcome. The SHAP dependency plot and force analysis provided model interpretation at the factor level and individual level, respectively.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e\u003cp\u003eWe have successfully developed, evaluated, and interpreted a machine learning model for predicting GN-BSI in ICU patients, facilitating timely interventions and treatments. The XGBoost model holds potential for clinical reference following validation set and further refinement.\u003c/p\u003e","manuscriptTitle":"A machine learning model for the early prediction of Gram-negative bloodstream infection in ICU patients","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-08 15:35:40","doi":"10.21203/rs.3.rs-8240564/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-01-14T13:54:47+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-01-02T18:03:16+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"309739985476436244043050287608595169569","date":"2025-12-06T18:15:34+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-06T10:54:45+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"121698190215656211834836396111958242755","date":"2025-12-04T22:08:17+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"222568325507061653835527463012656788901","date":"2025-12-04T18:11:11+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-12-04T18:00:00+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-12-04T09:08:52+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-12-03T10:22:42+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-12-03T10:20:34+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Medical Informatics and Decision Making","date":"2025-11-30T06:59:40+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-informatics-and-decision-making","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"midm","sideBox":"Learn more about [BMC Medical Informatics and Decision Making](http://bmcmedinformdecismak.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/midm/default.aspx","title":"BMC Medical Informatics and Decision Making","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"b3a84935-8388-4ca3-a0a5-6d40258319ca","owner":[],"postedDate":"December 8th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-02-02T16:00:37+00:00","versionOfRecord":{"articleIdentity":"rs-8240564","link":"https://doi.org/10.1186/s12911-026-03361-z","journal":{"identity":"bmc-medical-informatics-and-decision-making","isVorOnly":false,"title":"BMC Medical Informatics and Decision Making"},"publishedOn":"2026-01-31 15:58:11","publishedOnDateReadable":"January 31st, 2026"},"versionCreatedAt":"2025-12-08 15:35:40","video":"","vorDoi":"10.1186/s12911-026-03361-z","vorDoiUrl":"https://doi.org/10.1186/s12911-026-03361-z","workflowStages":[]},"version":"v1","identity":"rs-8240564","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8240564","identity":"rs-8240564","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}