Machine Learning Models for Predicting 28-Day Mortality in Gastrointestinal Bleeding with Acute Kidney Injury: A MIMIC-IV-Based Study

Machine Learning Models for Predicting 28-Day Mortality in Gastrointestinal Bleeding with Acute Kidney Injury: A MIMIC-IV-Based Study | 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 Machine Learning Models for Predicting 28-Day Mortality in Gastrointestinal Bleeding with Acute Kidney Injury: A MIMIC-IV-Based Study Xiangyu Zhang, Yanpeng Hu, Xingye Zhu, Chan Yu, Cuicui Liu, Jian Xue, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7339661/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 5 You are reading this latest preprint version Abstract Background: Gastrointestinal bleeding (GIB) is a common life-threatening condition in the digestive system that is frequently complicated by acute kidney injury (AKI), substantially increasing mortality and healthcare burden. To date, no precise tool exists for early prediction of short-term outcomes in patients with concurrent GIB and AKI (GIB-AKI).We conducted a retrospective cohort study aims to develop and validate machine learning (ML) models for predicting 28-day mortality . Methods: This retrospective cohort study was based on the MIMIC-IV database, including patients with first ICU admission who met criteria for GIB and AKI .From 64 clinical variables, we applied Least Absolute Shrinkage and Selection Operator (LASSO) regression and the Boruta algorithm to identify the key predictors of 28-day mortality. Five ML algorithms were employed to construct the prediction model. Model performance was evaluated based on identification, calibration, and clinical application.SHapley Additive exPlanations (SHAP) were used to interpret key variables influencing mortality in the best model. Results: A total of 1,890 adult GIB-AKI patients were included in this study. Five machine learning (ML) algorithms—Light Gradient Boosting Machine (LightGBM), Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), Logistic Regression, and Decision Tree—were developed and compared. Among all models, XGBoost demonstrated the best discriminative performance, achieving an area under the receiver operating characteristic curve (AUC) of 0.8027 in the validation set. Calibration and decision curve analyses confirmed its superior clinical utility. Furthermore, SHapley Additive exPlanations (SHAP)enhanced model interpretability by illustrating key variables influencing mortality. Conclusion: This study introduces a robust, interpretable ML model for early mortality prediction in GIB-AKI, offering a valuable tool for critical care management and precision medicine practice. Gastrointestinal bleeding Acute kidney injury Machine learning Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 1. Introduction Gastrointestinal bleeding (GIB) is a common life-threatening emergency of the digestive tract, encompassing both upper and lower gastrointestinal sources of hemorrhage[ 1 ]. Its clinical manifestations range from mild anemia to massive hemorrhage or even hypovolemic shock. Despite significant advances in endoscopic and interventional therapies in recent years, GIB remains a leading cause of increased hospital admissions and mortality[ 2 ]. Complications of GIB frequently involve multiple organ dysfunction, including respiratory failure, heart failure, and acute kidney injury (AKI)[ 3 ]. Previous studies have reported that AKI affects 25–30% of patients and is associated with increased mortality and prolonged hospital stays, thereby escalating healthcare burdens[ 4 – 5 ]. The pathophysiology of AKI in GIB is multifactorial, involving hypovolemia-induced renal hypoperfusion, nephrotoxic insults (e.g., NSAIDs and contrast media), and systemic inflammatory responses secondary to hemorrhage[ 6 – 9 ]. Early identification of GIB patients at high risk for AKI is therefore critical to guide individualized therapeutic and monitoring strategies. In clinical practice, prognostic tools such as the Glasgow-Blatchford Score (GBS), Rockall Score, and AIMS65 offer initial risk stratification for the general GIB population[ 10 ]. The GBS primarily guides decisions regarding hospital admission and urgent endoscopy, while the AIMS65 score, owing to its simplicity, provides a rapid estimate of short-term mortality risk. However, these tools do not account for the unique interplay of risk factors present in patients who develop AKI. Indeed, GIB-AKI patients commonly have multiple comorbidities, such as heart failure, cirrhosis, and infections, whose complex interactions challenge the predictive accuracy of traditional scoring systems and individualized risk assessment. With the rapid expansion of medical big data and artificial intelligence technologies, machine learning (ML) methods have shown promise in prognostic modeling for critically ill patients. ML algorithms can manage high-dimensional, nonlinear relationships and intricate variable interactions, often outperforming traditional statistical methods in mortality prediction and complication risk assessment[ 11 – 12 ]. In this context, we conducted a retrospective study using the MIMIC-IV database to develop and validate ML models for predicting 28-day all-cause mortality in GIB patients complicated by AKI. We extracted multidimensional clinical data, including demographics, comorbidities, vital signs, laboratory values, illness severity scores, and therapeutic interventions, and employed a dual feature-selection strategy [Least Absolute Shrinkage and Selection Operator (LASSO) and the Boruta algorithm]. Five ML algorithms were then constructed and compared: Light Gradient Boosting Machine (LightGBM), Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), Logistic Regression, and Decision Tree. This study aims to fill the prognostic gap in GIB-AKI by introducing a robust, interpretable risk-stratification tool that can assist clinicians in early risk identification and precision management in critical care settings. 2. Materials and Methods 2.1 Study Population This retrospective cohort study was conducted using data from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database, which contains clinical data from 383,220 intensive care unit (ICU) admissions at Beth Israel Deaconess Medical Center between 2008 and 2019. Access to the database was granted to the research team under user ID: 70784184. Inclusion criteria included: 1) documentation of GIB based on International Classification of Diseases (ICD) codes; 2) diagnosis of AKI according to KDIGO[ 13 ] criteria: urine output < 0.5 mL·kg⁻¹·h⁻¹ for ≥ 6 hours, increase in serum creatinine (SCr) ≥ 0.3 mg/dL within 48 hours, or SCr ≥ 1.5 times baseline within 7 days; 3) first ICU admission during the index hospitalization. Exclusion criteria were: 1) age < 18 years; 2) ICU stay < 24 hours; 3) multiple ICU admissions (only the first admission was retained); 4) preexisting end-stage renal disease. 2.2 Data Extraction and Feature Selection A total of 64 clinical variables were extracted, encompassing demographics, comorbidities, vital signs, laboratory values, severity scores, and treatments. Variables with > 20% missing values were excluded; patients with over 30% missingness across remaining variables were removed. For variables with ≤ 20% missing data, multiple imputation by chained equations (MICE) was performed in R. The cleaned dataset was then randomly divided using stratified sampling into a training set (70%, n = 1,323) and a validation set (30%, n = 567). Baseline characteristics between these two sets were compared using the Mann–Whitney U test for continuous variables and the chi-square test for categorical variables. To identify predictors of 28-day all-cause mortality, we first applied LASSO regression with 10-fold cross-validation to select the optimal penalty parameter (λ). Variables retained by LASSO were then subjected to the Boruta algorithm, which evaluates each feature’s importance by comparison with randomized “shadow” variables, yielding a final set of key predictors. 2.3 Model Construction and Evaluation Based on the selected variables, five ML algorithms were trained to estimate 28-day mortality risk: Light Gradient Boosting Machine (LightGBM), Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), Logistic Regression, and Decision Tree. Model performance was assessed in both training and validation cohorts. Receiver operating characteristic (ROC) curves were plotted, and the area under the curve (AUC) with 95% confidence intervals was calculated for each model. We further computed sensitivity, specificity, F₁-score, and accuracy to compare discriminative ability. Decision curve analysis (DCA) was performed to quantify net clinical benefit across a range of threshold probabilities.. The best-performing model was interpreted using SHapley Additive exPlanations (SHAP). Global feature importance was summarized via mean |SHAP| values, while individual predictions were elucidated with SHAP force plots, illustrating each feature’s positive or negative contribution to a given patient’s mortality risk. 2.5 Statistical Analysis Data were extracted using structured query language (SQL) via Navicat Premium (v15.0.12) with structured query language (SQL). Categorical variables are presented as medians with interquartile ranges and compared by the chi-square test. Continuous variables are reported as median (interquartile range) and compared by the Mann–Whitney U test. All statistical analyses were performed in R (v4.4.3). 3. Results 3.1 Baseline Characteristics After applying inclusion and exclusion criteria, 1,890 patients with GIB-AKI were included in the final cohort (684 females, 36.2%; 1,206 males, 63.8%). By day 28 of post-ICU admission, 452 had died, yielding an overall 28-day mortality rate of 23.9%. The patient selection flowchart is shown in Fig. 1 . The dataset was randomly divided into a training set (70%, n = 1,323) for model construction and a validation set (30%, n = 567) for model evaluation. As summarized in Table 1 , there were no statistically significant differences in baseline characteristics between the two cohorts (all p > 0.05), indicating good comparability. 3.2 Variable Selection To identify key variables associated with 28-day mortality, we employed a two-step feature selection strategy comprising LASSO regression followed by the Boruta algorithm, both applied to the training set. First, LASSO regression with 10-fold cross-validation identified 26 variables associated with 28-day mortality. The optimal regularization parameter was λ = 0.01462515 (log λ = − 4.225013) (Figs. 2 A,B). These variables are listed with their regression coefficients in Supplementary Table 1. Next, the Boruta algorithm was utilized to further analyze the 26 variables, assessing their importance. As shown in Fig. 2 C, 19 variables, highlighted in green, were retained as significant predictors of 28-day mortality. These 19 variables were incorporated into subsequent model construction and analysis. 3.3 Model Development and Performance Evaluation This study established five ML models to predict 28-day all-cause mortality in patients with GIB-AKI. In the training set, the AUC values for Logistic Regression, XGBoost, Decision Tree (DT), Support Vector Machine (SVM), and LightGBM models were 0.831, 0.882, 0.754, 0.844, and 0.848, respectively. The XGBoost model achieved the highest AUC (0.882, 95% CI: 0.862–0.903), as shown in Fig. 3 . In the validation set, the AUC values for Logistic Regression, XGBoost, Decision Tree (DT), Support Vector Machine (SVM), and LightGBM models were 0.785, 0.803, 0.700, 0.784, and 0.768, respectively. The XGBoost model demonstrated superior performance in the validation set with an AUC of 0.803 (95% CI: 0.759–0.846), as illustrated in Fig. 3 . Comparative performance metrics, including F₁-score, accuracy, precision, and recall, are summarized in Table 2 . XGBoost outperformed all other models across these key measures. Calibration curves (Figs. 4 ) d revealed that XGBoost predictions closely matched observed mortality rates in both training and validation sets, indicating a model fit exceeding that of the other four models. DCA analysis (Figs. 5 ) further confirmed that XGBoost yielded the greatest net clinical benefit over a wide range of threshold probabilities in predicting 28-day mortality compared to other models. 3.4 Interpretability Analysis Given its superior performance in both training and validation sets, the XGBoost model was subjected to SHapley Additive exPlanations (SHAP) analysis to elucidate feature contributions. Figure 6 A presents the SHAP summary plot of the top ten predictors ranked by mean |SHAP| value: APSIII score, AKI stage, Charlson Comorbidity Index, Age, Alkaline Phosphatase (ALP), Vasoactive drug use, Lactate, Diabetes, Anion gap, and Activated Partial Thromboplastin Time (APTT). In this plot, each point represents a single patient; its position on the x-axis reflects that feature’s SHAP value (positive values indicate increased mortality risk, while negative values indicate reduced risk), and color denotes the feature’s relative magnitude (yellow = high, purple = low). To illustrate individualized prediction, a SHAP force plot for one representative patient is shown in Fig. 6 B. Elevated APSIII score, higher AKI stage, increased bilirubin, raised PaCO₂ (reflecting respiratory compromise), elevated ALP, and larger mean corpuscular volume (MCV) are visualized as driving upward the patient’s predicted 28-day mortality risk. This level of transparency may aid clinicians in understanding and trusting the model’s outputs when making patient-specific decisions. 4. Discussion In this study, we developed and validated several ML models to predict 28-day all-cause mortality in patients with GIB-AKI, using data from the MIMIC-IV database. Among the five candidate models, the XGBoost model consistently outperformed traditional logistic regression and other ML models across multiple performance metrics, including discrimination, calibration, and clinical net benefit. The observed short-term mortality rate in the GIB-AKI cohort was 23.9%, indicating a poor overall prognosis and underscoring the need for accurate risk stratification tools. By combining LASSO regression and the Boruta algorithm for feature selection, we identified 19 key predictive variables. These variables—APSIII score, AKI stage, Charlson Comorbidity Index, lactate, alkaline phosphatase (ALP), and PaCO₂—are closely tied to illness severity and organ dysfunction, reflecting real-world clinical considerations. Among these predictors, the APSIII score emerged as the most important contributor. As a comprehensive index of disease severity in ICU patients, a higher APSIII score is strongly correlated with increased mortality risk[ 14 ], consistent with previous studies on prognostic factors for all-cause mortality in critically ill patients[ 15 – 16 ]. This finding reaffirms the central role of global physiological derangement in determining short-term outcomes in critically ill populations. AKI not only reflects acute deterioration of renal function but also often indicates systemic inflammatory response, hypovolemia, and tissue hypoxia. AKI stages also show a strong association with increased mortality risk, supporting its established role as an independent risk factor for ICU mortality[ 17 – 19 ]. The Charlson Index, a vital tool for measuring chronic comorbidity burden, shows that higher scores significantly increase both long-term and short-term mortality risks[ 20 ]. In this study, the Charlson Comorbidity Index, a well-validated measure of comorbid burden, was a key predictor. A higher Charlson score reflects increased vulnerability due to chronic diseases such as cardiovascular disorders, diabetes, cirrhosis, and malignancies, all of which are known to compromise physiologic reserve during acute stress and increase the risk of multi-organ failure[ 21 ]. Additionally, serum lactate was another major predictor. Lactate is a surrogate marker of tissue hypoperfusion and cellular metabolic stress and has been repeatedly linked to adverse outcomes in critically ill patients[ 22 – 23 ]. Its strong contribution to mortality prediction in our model aligns with prior evidence and emphasizes the importance of early identification of circulatory failure[ 24 – 25 ]. PaCO₂ was also among the top predictors. Elevated PaCO₂ may reflect hypoventilation, respiratory failure, or impaired compensatory mechanisms in the context of metabolic acidosis[ 26 – 27 ]. The positive association between PaCO₂ and mortality risk in our model highlights the prognostic importance of respiratory assessment in GIB-AKI patients. Traditional GIB risk scoring systems such as the Glasgow-Blatchford Score (GBS), AIMS65, and Rockall Score have proven useful for initial triage and early endoscopy decisions but are inadequate for capturing the complex risk profile of GIB patients complicated by AKI[ 28 ]. Our study introduces several key innovations: (1) Unlike general-purpose scores, our model is tailored specifically to GIB-AKI patients, resulting in improved precision and clinical relevance; 2) By leveraging a broader set of clinical variables and advanced ML techniques, our model captures complex, nonlinear relationships that are overlooked by traditional scoring methods; 3) The use of SHAP enhances transparency, allowing clinicians to understand how individual variables contribute to each prediction—an essential feature for clinical adoption. Our model exhibits strong clinical applicability. Early identification of high-risk GIB-AKI patients upon ICU admission can guide resource allocation, monitoring intensity, and individualized therapy. For instance, patients with high APSIII scores and elevated lactate may benefit from early organ support. Similarly, elevated PaCO₂ should prompt timely respiratory assessment and intervention. Identifying high-risk patients could facilitate early initiation of CRRT, aggressive resuscitation, and enhanced surveillance to mitigate clinical deterioration. Nevertheless, several limitations should be acknowledged. First, this is a single-center study based on the MIMIC-IV database, which may limit generalizability due to center-specific practices and patient populations. External validation in multicenter cohorts is essential. Second, although multiple imputation was employed to address missing data, residual bias from imputed variables cannot be fully excluded. Third, while SHAP enhances interpretability, it reflects statistical associations rather than causal relationships; thus, findings should be interpreted with clinical judgment. To enhance model robustness and applicability, future work should pursue 1) external validation across diverse healthcare systems; 2) incorporation of novel biomarkers, such as inflammatory cytokines or metabolomic profiles, alongside clinical variables; and 3) development of dynamic, time-updated prediction frameworks that integrate serial measurements to capture evolving patient trajectories. Ultimately, integrating such tools into electronic health records could enable real-time risk assessment, guiding precision interventions that improve outcomes for GIB-AKI patients. Conclusion We conducted a retrospective cohort study using data from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database, including 1,890 adult GIB-AKI patients. We applied Least Absolute Shrinkage and Selection Operator (LASSO) regression and the Boruta algorithm to identify 19 key predictors of 28-day mortality. Among the five machine learning (ML) algorithms, XGBoost demonstrated the best discriminative performance. Calibration and decision curve analyses confirmed its superior clinical utility. Furthermore, SHapley Additive exPlanations (SHAP) enhanced model interpretability by illustrating key variables influencing mortality. This study introduces a robust, interpretable ML model for early mortality prediction in GIB-AKI, offering a valuable tool for critical care management and precision medicine practice. Table 1 Baseline patient characteristics of GIB-AKI patients. Variables Training set(n = 1323) Validation set(n = 567) P value General information Age (years) 68.000 [56.000, 79.000] 67.000 [56.000, 79.000] 0.789 Gender,n(%) Female 475 (35.90) 209 (36.86) 0.730 Male 848 (64.10) 358 (63.14) Race,n(%) White 882 (66.67) 380 (67.02) 0.554 Black 126 (9.52) 65 (11.46) Asian 36 (2.72) 15 (2.65) Hispanic 65 (4.91) 21 (3.70) Other 214 (16.18) 86 (15.17) Weight(kg) 81.400 [68.750, 95.450] 80.400 [68.000, 95.700] 0.785 Los ICU(day) 3.260 [1.955, 6.625] 3.060 [1.930, 5.820] 0.292 Area of bleeding,n(%) Lower 266 (20.11) 109 (19.22) 0.787 Upper 830 (62.74) 354 (62.43) Unknown 227 (17.16) 104 (18.34) AKI stage,n(%) I 299 (22.60) 124 (21.87) 0.696 II 581 (43.92) 261 (46.03) III 443 (33.48) 182 (32.10) Vital signs HR (bpm) 92.000 [79.000, 106.000] 94.000 [80.000, 108.000] 0.155 SBP (mmHg) 118.000 [103.000, 137.000] 121.000 [105.000, 138.000] 0.142 DBP (mmHg) 65.000 [55.000, 77.000] 67.000 [56.000, 78.000] 0.268 MBP (mmHg) 79.000 [69.000, 92.000] 82.000 [70.000, 93.000] 0.208 RR (bpm) 20.000 [16.000, 24.000] 19.000 [16.000, 23.000] 0.132 Temperaturee(℃) 36.720 [36.440, 37.000] 36.720 [36.440, 37.060] 0.628 Urine output(mL) 1200.000 [741.500, 1865.000] 1215.000 [712.500, 1927.000] 0.995 SPO2(mmHg) 98.000 [95.000, 100.000] 98.000 [95.000, 100.000] 0.395 laboratory test Chloride(mmol/l) 105.000 [100.000, 109.000] 104.000 [99.000, 108.000] 0.076 Calcium(mmol/l) 8.100 [7.600, 8.700] 8.100 [7.600, 8.600] 0.707 Potassium(mmol/l) 4.200 [3.800, 4.700] 4.200 [3.800, 4.800] 0.857 Sodium(mmol/l) 139.000 [135.000, 142.000] 139.000 [134.500, 142.000] 0.113 Glucose (mg/dl) 128.000 [106.000, 169.000] 131.000 [106.000, 179.500] 0.305 Creatinine (mg/dL) 1.100 [0.800, 1.800] 1.200 [0.800, 1.900] 0.204 Aniongap(mmol/l) 14.000 [11.000, 18.000] 14.000 [12.000, 18.000] 0.297 BUN(mg/dL) 30.000 [18.500, 49.000] 31.000 [18.500, 54.500] 0.289 ALT (IU/L) 26.000 [15.000, 51.000] 26.000 [15.000, 54.500] 0.628 ALP (IU/L) 84.000 [59.000, 125.000] 81.000 [57.000, 124.500] 0.565 AST (IU/L) 45.000 [25.000, 99.000] 43.000 [24.000, 101.000] 0.717 Bilirubin(mg/dL) 1.100 [0.500, 3.000] 1.100 [0.500, 2.900] 0.594 WBC (×109/L) 10.700 [7.100, 15.800] 11.000 [7.550, 15.800] 0.600 RBC (×109/L) 3.020 [2.550, 3.585] 3.000 [2.560, 3.620] 0.815 PLT (×109/L) 155.000 [99.500, 238.000] 170.000 [105.500, 238.000] 0.291 HB(g/dL) 9.000 [7.700, 10.650] 9.000 [7.700, 10.700] 0.847 MCHC(g/L) 32.800 [31.600, 33.900] 32.900 [31.500, 34.000] 0.785 RDW(%) 16.000 [14.700, 18.000] 16.000 [14.650, 18.200] 0.717 MCV(fl) 92.000 [88.000, 97.000] 92.000 [88.000, 97.000] 0.678 MCH(pg) 30.200 [28.750, 31.900] 30.200 [28.500, 32.000] 0.706 HCT(%) 27.500 [23.600, 32.400] 27.700 [23.400, 32.200] 0.923 PT(s) 15.500 [13.400, 20.300] 15.500 [13.200, 19.600] 0.313 APTT(s) 31.600 [27.600, 39.050] 30.800 [26.800, 39.000] 0.182 INR 1.400 [1.200, 1.900] 1.400 [1.200, 1.800] 0.374 Lactate (mmol/L) 1.900 [1.300, 3.000] 2.000 [1.300, 3.150] 0.595 PO2(mmHg) 74.000 [43.000, 136.000] 70.000 [42.500, 137.000] 0.578 PCO2(mmHg) 40.000 [34.000, 47.000] 40.000 [35.000, 47.000] 0.842 PH 7.360 [7.280, 7.410] 7.350 [7.270, 7.410] 0.120 BE(mmol/L) -2.000 [-6.000, 0.000] -3.000 [-7.000, 0.000] 0.174 Comorbidity disease Myocardial infarct,n(%) No 1087 (82.16) 450 (79.37) 0.172 Yes 236 (17.84) 117 (20.63) Malignant cancer,n(%) No 1131 (85.49) 478 (84.30) 0.554 Yes 192 (14.51) 89 (15.70) Charlson comorbidity index 6.000 [4.000, 8.000] 6.000 [4.000, 8.000] 0.827 Severe liver disease,n(%) No 913 (69.01) 391 (68.96) 1.000 Yes 410 (30.99) 176 (31.04) Renal disease,n(%) No 973 (73.54) 424 (74.78) 0.615 Yes 350 (26.46) 143 (25.22) Congestive heart failure,n(%) No 899 (67.95) 392 (69.14) 0.651 Yes 424 (32.05) 175 (30.86) Peripheral vascular disease,n(%) No 1158 (87.53) 496 (87.48) 1.000 Yes 165 (12.47) 71 (12.52) Cerebrovascular disease,n(%) No 1178 (89.04) 504 (88.89) 0.987 Yes 145 (10.96) 63 (11.11) Chronic pulmonary disease,n(%) No 997 (75.36) 433 (76.37) 0.682 Yes 326 (24.64) 134 (23.63) Peptic ulcer disease,n(%) No 888 (67.12) 381 (67.20) 1.000 Yes 435 (32.88) 186 (32.80) Diabetes,n(%) No 917 (69.31) 384 (67.72) 0.530 Yes 406 (30.69) 183 (32.28) Treatment and severity scores Vasoactive,n(%) No 800 (60.47) 341 (60.14) 0.935 Yes 523 (39.53) 226 (39.86) SOFA (score) 5.00 [2.00, 8.00] 5.00 [2.00, 8.00] 0.938 GCS (score) 14.00 [12.00, 15.00] 14.0 [13.0, 15.0] 0.154 APSIII (score) 50.000 [40.000, 66.000] 52.000 [41.000, 67.000] 0.303 Ventilator,n(%) No 235 (17.76) 97 (17.11) 0.782 Yes 1088 (82.24) 470 (82.89) Endoscopy therpay,n(%) No 1221 (92.29) 530 (93.47) 0.419 Yes 102 (7.71) 37 (6.53) CRRT,n(%) No 1198 (90.55) 507 (89.42) 0.499 Yes 125 (9.45) 60 (10.58) Transfusion of red cells,n(%) No 1289 (97.43) 549 (96.83) 0.560 Yes 34 (2.57) 18 (3.17) Los ICU, length of ICU stay;HR, heart rate; SBP, systolic blood pressure; DBP, diastolic blood pressure; MAP, mean arterial pressure; RR, respiratory rate; BUN, blood urea nitrogen;ALT, alanine aminotransferase;AST, aspartate aminotransferase; ALP,alkaline phosphatase;RBC, red blood cell; WBC, white blood cell; PLT, platelet; RDW, erythrocyte distribution width; HB, hemoglobin; HCT, hematocrit;PT, prothrombin time; INR, international normalized ratio;BE,Base Excess;APTT, Activated Partial Thromboplastin Time;GCS, Glasgow coma scale; SOFA, sequential organ failure assessment; APSS III, Acute Physiology Score System III.CRRT, continuous renal replacement therapy; AKI, acute kidney injury. Table 2 Evaluation of five models performance in the training set and validation set. Model Accuracy Sensitivity Specificity F1 Training set lightGBM 0.810 0.769 0.823 0.659 Logistic Regression 0.752 0.832 0.727 0.616 SVM 0.822 0.794 0.831 0.681 Decision Tree 0.766 0.759 0.769 0.608 XGBoost 0.835 0.839 0.834 0.709 Validation set lightGBM 0.771 0.654 0.807 0.578 Logistic Regression 0.732 0.787 0.715 0.585 SVM 0.771 0.706 0.791 0.596 Decision Tree 0.757 0.706 0.773 0.582 XGBoost 0.790 0.765 0.798 0.636 Abbreviations Los ICU, length of ICU stay;HR, heart rate; SBP, systolic blood pressure; DBP, diastolic blood pressure; MAP, mean arterial pressure; RR, respiratory rate; BUN, blood urea nitrogen;ALT, alanine aminotransferase;AST, aspartate aminotransferase; ALP,alkaline phosphatase;RBC, red blood cell; WBC, white blood cell; PLT, platelet; RDW, erythrocyte distribution width; HB, hemoglobin; HCT, hematocrit;PT, prothrombin time; INR, international normalized ratio;BE,Base Excess;APTT, Activated Partial Thromboplastin Time;GCS, Glasgow coma scale; SOFA, sequential organ failure assessment; APSS III, Acute Physiology Score System III.CRRT, continuous renal replacement therapy; AKI, acute kidney injury. Declarations Ethics approval and consent to participate The datasets generated and analysed during the current study are available in the MIMIC-IV (v. 3.1) database, which contains deidentified data. The database complies with the ethical guidelines outlined in the Declaration of Helsinki. Therefore, additional approval from the institutional review board was not required. Access to the database was granted to the research team under user ID: 70784184. Consent for publication Not applicable. Availability of data and materials The data presented in the current study are available in the MIMIC-IV database (https://physionet.org/content/mimiciv/). Competing interests The authors declare no competing interests. Funding No funding support. Authors and Affiliations Department of Gastrointestinal Surgery, Qilu Hospital of Shandong University Dezhou Hospital, 1751 Xinhu Street, Dezhou 253000, China Xiangyu Zhang,Yanpeng Hu,Xingye Zhu,Chan Yu,Cuicui Liu,Jian Xue,Yingfeng Su,Baoqing Ma. Xiangyu Zhang and Yanpeng Hu have contributed equally to this work and share the first authorship. Author Contribution Conceptualization: Xiangyu Zhang, Baoqing Ma. Data curation: Xiangyu Zhang, Cuicui Liu, Yanpeng Hu. Formal analysis: Yanpeng Hu, Xingye Zhu, Jian Xue. Writing—original draft: Xiangyu Zhang, Baoqing Ma, Yingfeng Su. Writing—review and editing: Xingye Zhu, Chan Yu,Baoqing Ma. Corresponding author Correspondence to Baoqing Ma. Acknowledgements We are grateful to the MIMIV-IV participants and staff. We appreciate all the reviewers who participated in the review. References Long B, Gottlieb M. Emergency medicine updates: Upper gastrointestinal bleeding. Am J Emerg Med. 2024;81:116-123. Hong MJ,Lee SY,Kim JH, et al. Rebleeding after initial endoscopic hemostasis in peptic ulcer disease. J Korean Med Sci. 2014;29 (10):1411-5. Morarasu BC, Sorodoc V, Haisan A, et al. Age, blood tests and comorbidities and AIMS65 risk scores outperform Glasgow-Blatchford and pre-endoscopic Rockall score in patients with upper gastrointestinal bleeding. World J Clin Cases. 2023;11(19):4513-4530. Zhong L,Quan X,Dang P, et al. Clinical characteristics and risk factors of in-hospital gastrointestinal bleeding in patients with acute myocardial infarction. Front Cardiovasc Med. 2022;9:933597. Cakmak U, Merhametsiz O, Gok Oguz E, et al. Effects of acute kidney injury on clinical outcomes in patients with upper gastrointestinal bleeding. Ren Fail. 2016;38(2):176-184. Zhang H,Wang Z,Tang Y, et al. Prediction of acute kidney injury after cardiac surgery: model development using a Chinese electronic health record dataset. J Transl Med. 2022;20 (1):166. Castellano G,Franzin R,Sallustio F, et al. Complement component C5a induces aberrant epigenetic modifications in renal tubular epithelial cells accelerating senescence by Wnt4/βcatenin signaling after ischemia/reperfusion injury. Aging (Albany NY). 2019;11 (13):4382-4406. Kwak MS,Cha JM,Han YJ, et al. The Clinical Outcomes of Lower Gastrointestinal Bleeding Are Not Better than Those of Upper Gastrointestinal Bleeding. J Korean Med Sci. 2016;31 (10):1611-6. Wan X,Xie X,Gendoo Y, et al. Ulinastatin administration is associated with a lower incidence of acute kidney injury after cardiac surgery: a propensity score matched study. Crit Care. 2016;20:42. Tantai XX,Liu N,Yang LB, et al. Prognostic value of risk scoring systems for cirrhotic patients with variceal bleeding. World J Gastroenterol. 2019;25 (45):6668-6680. Assaf D,Gutman Y,Neuman Y, et al. Utilization of machine-learning models to accurately predict the risk for critical COVID-19. Intern Emerg Med. 2020;15 (8):1435-1443. Explainable Machine-Learning Model for Prediction of In-Hospital Mortality in Septic Patients Requiring Intensive Care Unit Readmission. Infect Dis Ther. 2022;11 (4):1695-1713. Meersch M,Schmidt C,Hoffmeier A, et al. Prevention of cardiac surgery-associated AKI by implementing the KDIGO guidelines in high risk patients identified by biomarkers: the PrevAKI randomized controlled trial. Intensive Care Med. 2017;43 (11):1551-1561. Xu Y,Han D,Huang T, et al. Predicting ICU Mortality in Rheumatic Heart Disease: Comparison of XGBoost and Logistic Regression. Front Cardiovasc Med. 2022;9:847206. Langelier C,Kalantar KL,Moazed F, et al. Integrating host response and unbiased microbe detection for lower respiratory tract infection diagnosis in critically ill adults. Proc Natl Acad Sci U S A. 2018;115 (52):E12353-E12362. Xue G,Liang H,Ye J, et al. Development and Validation of a Predictive Scoring System for In-hospital Death in Patients With Intra-Abdominal Infection: A Single-Center 10-Year Retrospective Study. Front Med (Lausanne). 2021;8:741914. Hosseini-Moghaddam SM,Luo B,Bota SE, et al. Incidence and Outcomes Associated With Clostridioides difficile Infection in Solid Organ Transplant Recipients. JAMA Netw Open. 2021;4 (12):e2141089. Yu WQ,Zhang SY,Fu SQ, et al. Dexamethasone protects the glycocalyx on the kidney microvascular endothelium during severe acute pancreatitis. J Zhejiang Univ Sci B. 2019;20 (4):355-362. Haase-Fielitz A,Altendeitering F,Iwers R, et al. Acute kidney injury may impede results after transcatheter aortic valve implantation. Clin Kidney J. 2021;14 (1):261-268. Kallogjeri D,Gaynor SM,Piccirillo ML, et al. Comparison of comorbidity collection methods. J Am Coll Surg. 2014;219 (2):245-55. Yu Q,Wang Y,Huang S, et al. Multicenter cohort study demonstrates more consolidation in upper lungs on initial CT increases the risk of adverse clinical outcome in COVID-19 patients. Theranostics. 2020;10 (12):5641-5648. Schierenbeck F,Nijsten MW,Franco-Cereceda A, et al. Introducing intravascular microdialysis for continuous lactate monitoring in patients undergoing cardiac surgery: a prospective observational study. Crit Care. 2014;18 (2):R56. Gao F,Huang XL,Cai MX, et al. Prognostic value of serum lactate kinetics in critically ill patients with cirrhosis and acute-on-chronic liver failure: a multicenter study. Aging (Albany NY). 2019;11 (13):4446-4462. Daga MK,Rohatgi I,Mishra R, et al. Lactate enhanced-quick Sequential Organ Failure Assessment 2 (LqSOFA2): A new score for bedside prognostication of patients with sepsis. Indian J Med Res. 2021;154 (4):607-614. Caruso V,Besch G,Nguyen M, et al. Treatment of Hyperlactatemia in Acute Circulatory Failure Based on CO 2 -O 2 -Derived Indices: Study Protocol for a Prospective, Multicentric, Single, Blind, Randomized, Superiority Study (The LACTEL Study). Front Cardiovasc Med. 2022;9:898406. Lee SH,Park S,Lee JW, et al. The Anion Gap is a Predictive Clinical Marker for Death in Patients with Acute Pesticide Intoxication. J Korean Med Sci. 2016;31 (7):1150-9. Robin E,Futier E,Pires O, et al. Central venous-to-arterial carbon dioxide difference as a prognostic tool in high-risk surgical patients. Crit Care. 2015;19:227. Shi H, Shen Y, Li L. Early prediction of acute kidney injury in patients with gastrointestinal bleeding admitted to the intensive care unit based on extreme gradient boosting. Front Med (Lausanne). 2023;10:1221602. Additional Declarations No competing interests reported. Supplementary Files supplementTable1.xls Cite Share Download PDF Status: Under Review Version 1 posted Reviewers invited by journal 19 Sep, 2025 Editor invited by journal 01 Sep, 2025 Editor assigned by journal 25 Aug, 2025 Submission checks completed at journal 23 Aug, 2025 First submitted to journal 23 Aug, 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-7339661","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":522215818,"identity":"e9d62dc1-10e5-4e61-8be0-1898482820ad","order_by":0,"name":"Xiangyu Zhang","email":"","orcid":"","institution":"Qilu Hospital of Shandong University Dezhou Hospital","correspondingAuthor":false,"prefix":"","firstName":"Xiangyu","middleName":"","lastName":"Zhang","suffix":""},{"id":522215821,"identity":"bcc84480-0945-4e56-94a7-8550e58e0c4a","order_by":1,"name":"Yanpeng Hu","email":"","orcid":"","institution":"Qilu Hospital of Shandong University Dezhou Hospital","correspondingAuthor":false,"prefix":"","firstName":"Yanpeng","middleName":"","lastName":"Hu","suffix":""},{"id":522215822,"identity":"aa0db0cf-194e-4364-a72f-9c4d06cf21fe","order_by":2,"name":"Xingye Zhu","email":"","orcid":"","institution":"Qilu Hospital of Shandong University Dezhou Hospital","correspondingAuthor":false,"prefix":"","firstName":"Xingye","middleName":"","lastName":"Zhu","suffix":""},{"id":522215823,"identity":"29c216e9-e301-4e33-b1e2-48b2ef44fd84","order_by":3,"name":"Chan Yu","email":"","orcid":"","institution":"Qilu Hospital of Shandong University Dezhou Hospital","correspondingAuthor":false,"prefix":"","firstName":"Chan","middleName":"","lastName":"Yu","suffix":""},{"id":522215825,"identity":"f42329fc-aae1-484a-b638-6aaab4a722d3","order_by":4,"name":"Cuicui Liu","email":"","orcid":"","institution":"Qilu Hospital of Shandong University Dezhou Hospital","correspondingAuthor":false,"prefix":"","firstName":"Cuicui","middleName":"","lastName":"Liu","suffix":""},{"id":522215826,"identity":"296fb7b9-7930-4574-84de-3e9b6fb271eb","order_by":5,"name":"Jian Xue","email":"","orcid":"","institution":"Qilu Hospital of Shandong University Dezhou Hospital","correspondingAuthor":false,"prefix":"","firstName":"Jian","middleName":"","lastName":"Xue","suffix":""},{"id":522215827,"identity":"908bf89a-fde0-454d-97fb-27fbe8320392","order_by":6,"name":"Yingfeng Su","email":"","orcid":"","institution":"Qilu Hospital of Shandong University Dezhou Hospital","correspondingAuthor":false,"prefix":"","firstName":"Yingfeng","middleName":"","lastName":"Su","suffix":""},{"id":522215828,"identity":"a9d2a8fe-101d-496e-a3e2-7bc849dc4924","order_by":7,"name":"Baoqing Ma","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA6ElEQVRIiWNgGAWjYFACHhAhAcTnHz74YGBjR4IWxjPMhjMK0pKJ1QIEzGfYhHk+HGJsIKRBt/3swQ8fyizy5N3OHmO2MTjAzMB++OgGfFrMzuQlS844J1FseOZc2uMcgzt8DDxpaTfwajmQYyDN2yaRuHHGAXPjHINnzAwSPGb4tZx/Y/wbrGX+AzNpC4PDjA0EtdzIMQPbMp/hjJk0A3Fa3qVZAv2SuIHhWLJhj0FaMhtBv5zPPXzjQ1ld4vyGwwcf/PhjY8fPfvgYXi0QwMbAYHAAwSYGAJXJNxCndBSMglEwCkYgAAAh81EGiDz+0QAAAABJRU5ErkJggg==","orcid":"","institution":"Qilu Hospital of Shandong University Dezhou Hospital","correspondingAuthor":true,"prefix":"","firstName":"Baoqing","middleName":"","lastName":"Ma","suffix":""}],"badges":[],"createdAt":"2025-08-10 15:23:09","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7339661/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7339661/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":92583455,"identity":"941f5ab1-03d1-407b-8aab-145b24f0203c","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"jpg","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":93892,"visible":true,"origin":"","legend":"","description":"","filename":"figure1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/8f6097bd194b7bbd77a20035.jpg"},{"id":92584691,"identity":"f15cf6cd-1b6c-4335-a5f6-15b382bafd77","added_by":"auto","created_at":"2025-10-01 10:15:46","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":444749,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/1ade19b2059a3e6bc87556cd.docx"},{"id":92583457,"identity":"55076b14-61fd-420b-bbdb-1b4ec7f3ffa7","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"jpg","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4360019,"visible":true,"origin":"","legend":"","description":"","filename":"figure2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/47bb1851181133164dd298f8.jpg"},{"id":92583461,"identity":"e376a9e5-8bb4-4960-8274-af7d71e70af4","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"jpg","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4101819,"visible":true,"origin":"","legend":"","description":"","filename":"figure3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/4bd55c8d3dc4994bb4dd650e.jpg"},{"id":92585681,"identity":"bbbe559d-9487-4eec-bee9-daea3f26b212","added_by":"auto","created_at":"2025-10-01 10:31:46","extension":"csv","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5232,"visible":true,"origin":"","legend":"","description":"","filename":"Table1.csv","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/2d1180a35c3b2d8b19b51c2f.csv"},{"id":92583476,"identity":"5786ccc9-75a8-42e3-b877-aab444b3eb67","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"jpg","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3726767,"visible":true,"origin":"","legend":"","description":"","filename":"figure4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/7668d9dfc7125e79f8da6aa5.jpg"},{"id":92583467,"identity":"463ff668-f19c-4c71-85ef-d9dc1b31fdc5","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"xls","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19968,"visible":true,"origin":"","legend":"","description":"","filename":"Table2.xls","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/71f58fc3693924021a47780d.xls"},{"id":92583473,"identity":"978d41e3-e5cd-4183-bbef-edaf7a823383","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"jpg","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3475452,"visible":true,"origin":"","legend":"","description":"","filename":"figure5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/439549ae6b38319e00244e56.jpg"},{"id":92584696,"identity":"327232e9-9dcc-430b-bf81-6b3e6015ae94","added_by":"auto","created_at":"2025-10-01 10:15:46","extension":"jpg","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3155939,"visible":true,"origin":"","legend":"","description":"","filename":"figure6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/fc13c97bcce9e24415304348.jpg"},{"id":92583471,"identity":"d6dedc44-a0c5-4f93-9ba4-9890dd330c72","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"json","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9379,"visible":true,"origin":"","legend":"","description":"","filename":"dbec457bba46465bab9bc16f6bc75a71.json","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/de767400b17f6302e8dfb84e.json"},{"id":92583474,"identity":"4aa31586-53df-439c-bf9e-a4b67b8a21c2","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"xls","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":21504,"visible":true,"origin":"","legend":"","description":"","filename":"supplementTable1.xls","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/51043fa9ae20bb51edcf9116.xls"},{"id":92584697,"identity":"03b2d91b-426b-4ad3-b6d5-0b7c3d35b5de","added_by":"auto","created_at":"2025-10-01 10:15:46","extension":"xml","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":120616,"visible":true,"origin":"","legend":"","description":"","filename":"dbec457bba46465bab9bc16f6bc75a711enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/a8f0516abb72d8c11b0452cd.xml"},{"id":92583470,"identity":"f27258e4-b632-43ba-b4b3-9968ff39ec35","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"jpg","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":93892,"visible":true,"origin":"","legend":"","description":"","filename":"figure1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/f1df808948f600a9ea3077b5.jpg"},{"id":92583475,"identity":"95cf91b3-7e91-43b9-9ef1-652537e9815f","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"jpg","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4360019,"visible":true,"origin":"","legend":"","description":"","filename":"figure2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/7bf88e1cc9c486a9a1018463.jpg"},{"id":92583495,"identity":"b5b11b44-7e86-482f-8b5b-624e67c872c3","added_by":"auto","created_at":"2025-10-01 10:07:47","extension":"jpg","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4101819,"visible":true,"origin":"","legend":"","description":"","filename":"figure3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/6658b4e0a77c0ce82616775b.jpg"},{"id":92585411,"identity":"0bc150ea-8cc7-4598-931c-17cf9e835428","added_by":"auto","created_at":"2025-10-01 10:23:46","extension":"jpg","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3726767,"visible":true,"origin":"","legend":"","description":"","filename":"figure4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/52e746e1aaa32ad1a0172c11.jpg"},{"id":92583500,"identity":"55918590-f2e2-4121-bc1a-84994c63c1d5","added_by":"auto","created_at":"2025-10-01 10:07:47","extension":"jpg","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3475452,"visible":true,"origin":"","legend":"","description":"","filename":"figure5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/825cf7ea531d9840cc726b26.jpg"},{"id":92585412,"identity":"054c1c0f-e672-4578-bc84-51da7a7f315b","added_by":"auto","created_at":"2025-10-01 10:23:46","extension":"jpg","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3155939,"visible":true,"origin":"","legend":"","description":"","filename":"figure6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/acf9388a754ea9d26a6450c1.jpg"},{"id":92583472,"identity":"ad102e78-db2e-42ea-92ac-e673e88a4f5b","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"jpeg","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":350763,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/86bc1db79e514e5f184aa4b2.jpeg"},{"id":92583485,"identity":"c9ca9d3e-f449-4da7-9636-6e8567a2f26d","added_by":"auto","created_at":"2025-10-01 10:07:47","extension":"jpeg","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":224495,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/4ecd17bc51e3d5645141bfd5.jpeg"},{"id":92583487,"identity":"76f926e8-b70a-4b65-8a53-86e7308fa8e0","added_by":"auto","created_at":"2025-10-01 10:07:47","extension":"jpeg","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":372735,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/2507949505110cc85c15324c.jpeg"},{"id":92585413,"identity":"4aefc417-776d-4d7f-a5de-4dc62b067112","added_by":"auto","created_at":"2025-10-01 10:23:46","extension":"jpeg","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":267266,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/da73bf560da96dcefe7406fb.jpeg"},{"id":92583492,"identity":"6eeb0309-c3a1-4e40-9e79-dc93efdff736","added_by":"auto","created_at":"2025-10-01 10:07:47","extension":"jpeg","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":233707,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/03402165623996552ebd02fd.jpeg"},{"id":92583478,"identity":"698a0a14-04f1-4227-b179-2f75fbf48d36","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"jpeg","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":213260,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/b2c44babb6a21072f99ff5f1.jpeg"},{"id":92584698,"identity":"30f264cf-491a-462f-8a99-15b0f65319a4","added_by":"auto","created_at":"2025-10-01 10:15:46","extension":"png","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":31469,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefigure1.png","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/daf8038694966d86ca0011e1.png"},{"id":92583498,"identity":"acf3ddaa-e4eb-438f-860e-a6b97d2b666b","added_by":"auto","created_at":"2025-10-01 10:07:47","extension":"png","order_by":25,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":348619,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefigure2.png","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/f33f7d21f70763aa984b31f6.png"},{"id":92584695,"identity":"0c203636-2a55-40a8-859f-f90979c094a6","added_by":"auto","created_at":"2025-10-01 10:15:46","extension":"png","order_by":26,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":326772,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefigure3.png","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/9fe500452fc1fb7e3b0fb4f7.png"},{"id":92583493,"identity":"521dd4c6-f63f-4ded-b016-dfb097d9c20d","added_by":"auto","created_at":"2025-10-01 10:07:47","extension":"png","order_by":27,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":386198,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefigure4.png","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/b335aec38c8cf2c489fd787c.png"},{"id":92583494,"identity":"482b456a-99d1-4dc9-b488-b50548d17b18","added_by":"auto","created_at":"2025-10-01 10:07:47","extension":"png","order_by":28,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":245572,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefigure5.png","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/3153c6a161d25ed00366bc49.png"},{"id":92584701,"identity":"d74bcb54-4077-4a5b-80c3-76ef8e7bb12a","added_by":"auto","created_at":"2025-10-01 10:15:47","extension":"png","order_by":29,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":370963,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefigure6.png","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/5c14b3cba8bc9c2b7468d492.png"},{"id":92583477,"identity":"91f5e249-7038-4ad1-8d8f-4df41c58ed55","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"png","order_by":30,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":61877,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/961d832aadc902edc0627125.png"},{"id":92583499,"identity":"26da19fa-b287-405f-8bcd-c120ea70360d","added_by":"auto","created_at":"2025-10-01 10:07:47","extension":"png","order_by":31,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":44567,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/e94f101e0a015c2c13b9da74.png"},{"id":92585415,"identity":"cc05d7c3-017e-416b-9df3-e5bebc05c70f","added_by":"auto","created_at":"2025-10-01 10:23:47","extension":"png","order_by":32,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":64717,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/e320bbc2ce268fd1062d5cfd.png"},{"id":92583484,"identity":"c1868a73-1248-4c8e-9ce1-07f69b3fb2ab","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"png","order_by":33,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":45329,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/332b32e8318fd65685d3e8bb.png"},{"id":92583479,"identity":"53e6b8b7-daf7-4e06-a071-60351f8abe00","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"png","order_by":34,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":44826,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/f24d8e653ee65b4e8679becc.png"},{"id":92583482,"identity":"77b710f6-6dce-4626-8b12-0cfd5e8211e7","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"png","order_by":35,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":49110,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/fa7bc44fbe72335a4881670a.png"},{"id":92585682,"identity":"debe4ef4-7deb-4176-a880-db04380d388d","added_by":"auto","created_at":"2025-10-01 10:31:46","extension":"xml","order_by":36,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":116261,"visible":true,"origin":"","legend":"","description":"","filename":"dbec457bba46465bab9bc16f6bc75a711structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/43f3ada4295110a55a94e5a3.xml"},{"id":92584703,"identity":"25dcbdc8-889b-48a0-bc7f-3cda58b8f132","added_by":"auto","created_at":"2025-10-01 10:15:47","extension":"html","order_by":37,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":127041,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/a28cd5a48892d39fd1a30162.html"},{"id":92585408,"identity":"03aee5a1-6e68-4d84-8c64-3902a6648baa","added_by":"auto","created_at":"2025-10-01 10:23:46","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":93892,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart illustrating the GIB-AKI patient selection process.\u003c/p\u003e","description":"","filename":"figure1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/d3106bde9fe539474d38d81a.jpg"},{"id":92584686,"identity":"ca43ff4f-2ffe-4a92-b888-4f8f048c2b31","added_by":"auto","created_at":"2025-10-01 10:15:46","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":4360019,"visible":true,"origin":"","legend":"\u003cp\u003eLASSO parameter selection and coefficient profiles. (A) The optimal tuning parameter selection map for the LASSO analysis. (B) Feature selection using the Boruta algorithm.(C)\u003c/p\u003e","description":"","filename":"figure2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/2fae4dbfeba954f51f3dfd57.jpg"},{"id":92584690,"identity":"e34ef086-ebac-4aae-a17c-3d6970696767","added_by":"auto","created_at":"2025-10-01 10:15:46","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":4101819,"visible":true,"origin":"","legend":"\u003cp\u003eThe Receiver operating characteristic(ROC) curves comparison of the five models in training set(A) and validation set(B).\u003c/p\u003e","description":"","filename":"figure3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/587ca44da8a51142a19b6115.jpg"},{"id":92583463,"identity":"b2fa3522-b2c5-4045-a040-b2ed7e9b0dbc","added_by":"auto","created_at":"2025-10-01 10:07:46","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":3726767,"visible":true,"origin":"","legend":"\u003cp\u003eThe Calibration curves of the five models in training set(A) and validation set(B).\u003c/p\u003e","description":"","filename":"figure4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/47267d626022d69ccc292a76.jpg"},{"id":92584692,"identity":"0346cd5d-8446-425c-b68c-2b488ff50c32","added_by":"auto","created_at":"2025-10-01 10:15:46","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":3475452,"visible":true,"origin":"","legend":"\u003cp\u003eThe decision curve analysis (DCA) curve of the five models in training set(A) and validation set(B).\u003c/p\u003e","description":"","filename":"figure5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/2f1c4ceb3d5727dffab285ea.jpg"},{"id":92584693,"identity":"6282408a-aecb-4d2f-8f50-14f4d70188d1","added_by":"auto","created_at":"2025-10-01 10:15:46","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":3155939,"visible":true,"origin":"","legend":"\u003cp\u003eSHAP (Shapley additive explanations) analysis of the top-10 predictors for 28-Day Mortality in GIB-AKI patients using a XGBoost model ranked by mean absolute SHAP value.(A) A SHAP force plot for one representative patient interpret machine learning models visually.(B)\u003c/p\u003e","description":"","filename":"figure6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/d34f5d153c3532bc6a80e031.jpg"},{"id":92586318,"identity":"cc85e016-2007-4c0f-a456-c2eb40c73f44","added_by":"auto","created_at":"2025-10-01 10:39:53","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":19875316,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/80966f16-d90a-4cb6-a7c9-e94d099bd99d.pdf"},{"id":92585409,"identity":"613f5909-6932-467f-b109-202bd868f784","added_by":"auto","created_at":"2025-10-01 10:23:46","extension":"xls","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":21504,"visible":true,"origin":"","legend":"","description":"","filename":"supplementTable1.xls","url":"https://assets-eu.researchsquare.com/files/rs-7339661/v1/e2b8c973db57bb7ea6584f16.xls"}],"financialInterests":"No competing interests reported.","formattedTitle":"Machine Learning Models for Predicting 28-Day Mortality in Gastrointestinal Bleeding with Acute Kidney Injury: A MIMIC-IV-Based Study","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eGastrointestinal bleeding (GIB) is a common life-threatening emergency of the digestive tract, encompassing both upper and lower gastrointestinal sources of hemorrhage[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Its clinical manifestations range from mild anemia to massive hemorrhage or even hypovolemic shock. Despite significant advances in endoscopic and interventional therapies in recent years, GIB remains a leading cause of increased hospital admissions and mortality[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Complications of GIB frequently involve multiple organ dysfunction, including respiratory failure, heart failure, and acute kidney injury (AKI)[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Previous studies have reported that AKI affects 25\u0026ndash;30% of patients and is associated with increased mortality and prolonged hospital stays, thereby escalating healthcare burdens[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. The pathophysiology of AKI in GIB is multifactorial, involving hypovolemia-induced renal hypoperfusion, nephrotoxic insults (e.g., NSAIDs and contrast media), and systemic inflammatory responses secondary to hemorrhage[\u003cspan additionalcitationids=\"CR7 CR8\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Early identification of GIB patients at high risk for AKI is therefore critical to guide individualized therapeutic and monitoring strategies.\u003c/p\u003e\u003cp\u003eIn clinical practice, prognostic tools such as the Glasgow-Blatchford Score (GBS), Rockall Score, and AIMS65 offer initial risk stratification for the general GIB population[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. The GBS primarily guides decisions regarding hospital admission and urgent endoscopy, while the AIMS65 score, owing to its simplicity, provides a rapid estimate of short-term mortality risk. However, these tools do not account for the unique interplay of risk factors present in patients who develop AKI. Indeed, GIB-AKI patients commonly have multiple comorbidities, such as heart failure, cirrhosis, and infections, whose complex interactions challenge the predictive accuracy of traditional scoring systems and individualized risk assessment.\u003c/p\u003e\u003cp\u003eWith the rapid expansion of medical big data and artificial intelligence technologies, machine learning (ML) methods have shown promise in prognostic modeling for critically ill patients. ML algorithms can manage high-dimensional, nonlinear relationships and intricate variable interactions, often outperforming traditional statistical methods in mortality prediction and complication risk assessment[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eIn this context, we conducted a retrospective study using the MIMIC-IV database to develop and validate ML models for predicting 28-day all-cause mortality in GIB patients complicated by AKI. We extracted multidimensional clinical data, including demographics, comorbidities, vital signs, laboratory values, illness severity scores, and therapeutic interventions, and employed a dual feature-selection strategy [Least Absolute Shrinkage and Selection Operator (LASSO) and the Boruta algorithm]. Five ML algorithms were then constructed and compared: Light Gradient Boosting Machine (LightGBM), Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), Logistic Regression, and Decision Tree. This study aims to fill the prognostic gap in GIB-AKI by introducing a robust, interpretable risk-stratification tool that can assist clinicians in early risk identification and precision management in critical care settings.\u003c/p\u003e"},{"header":"2. Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1 Study Population\u003c/h2\u003e\u003cp\u003eThis retrospective cohort study was conducted using data from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database, which contains clinical data from 383,220 intensive care unit (ICU) admissions at Beth Israel Deaconess Medical Center between 2008 and 2019. Access to the database was granted to the research team under user ID: 70784184.\u003c/p\u003e\u003cp\u003eInclusion criteria included: 1) documentation of GIB based on International Classification of Diseases (ICD) codes; 2) diagnosis of AKI according to KDIGO[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e] criteria: urine output\u0026thinsp;\u0026lt;\u0026thinsp;0.5 mL\u0026middot;kg⁻\u0026sup1;\u0026middot;h⁻\u0026sup1; for \u0026ge;\u0026thinsp;6 hours, increase in serum creatinine (SCr)\u0026thinsp;\u0026ge;\u0026thinsp;0.3 mg/dL within 48 hours, or SCr\u0026thinsp;\u0026ge;\u0026thinsp;1.5 times baseline within 7 days; 3) first ICU admission during the index hospitalization.\u003c/p\u003e\u003cp\u003eExclusion criteria were: 1) age\u0026thinsp;\u0026lt;\u0026thinsp;18 years; 2) ICU stay\u0026thinsp;\u0026lt;\u0026thinsp;24 hours; 3) multiple ICU admissions (only the first admission was retained); 4) preexisting end-stage renal disease.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2 Data Extraction and Feature Selection\u003c/h2\u003e\u003cp\u003eA total of 64 clinical variables were extracted, encompassing demographics, comorbidities, vital signs, laboratory values, severity scores, and treatments. Variables with \u0026gt;\u0026thinsp;20% missing values were excluded; patients with over 30% missingness across remaining variables were removed. For variables with \u0026le;\u0026thinsp;20% missing data, multiple imputation by chained equations (MICE) was performed in R. The cleaned dataset was then randomly divided using stratified sampling into a training set (70%, n\u0026thinsp;=\u0026thinsp;1,323) and a validation set (30%, n\u0026thinsp;=\u0026thinsp;567). Baseline characteristics between these two sets were compared using the Mann\u0026ndash;Whitney U test for continuous variables and the chi-square test for categorical variables.\u003c/p\u003e\u003cp\u003eTo identify predictors of 28-day all-cause mortality, we first applied LASSO regression with 10-fold cross-validation to select the optimal penalty parameter (λ). Variables retained by LASSO were then subjected to the Boruta algorithm, which evaluates each feature\u0026rsquo;s importance by comparison with randomized \u0026ldquo;shadow\u0026rdquo; variables, yielding a final set of key predictors.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e2.3 Model Construction and Evaluation\u003c/h2\u003e\u003cp\u003eBased on the selected variables, five ML algorithms were trained to estimate 28-day mortality risk: Light Gradient Boosting Machine (LightGBM), Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), Logistic Regression, and Decision Tree. Model performance was assessed in both training and validation cohorts. Receiver operating characteristic (ROC) curves were plotted, and the area under the curve (AUC) with 95% confidence intervals was calculated for each model. We further computed sensitivity, specificity, F₁-score, and accuracy to compare discriminative ability. Decision curve analysis (DCA) was performed to quantify net clinical benefit across a range of threshold probabilities..\u003c/p\u003e\u003cp\u003eThe best-performing model was interpreted using SHapley Additive exPlanations (SHAP). Global feature importance was summarized \u003cem\u003evia\u003c/em\u003e mean |SHAP| values, while individual predictions were elucidated with SHAP force plots, illustrating each feature\u0026rsquo;s positive or negative contribution to a given patient\u0026rsquo;s mortality risk.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.5 Statistical Analysis\u003c/h2\u003e\u003cp\u003eData were extracted using structured query language (SQL) \u003cem\u003evia\u003c/em\u003e Navicat Premium (v15.0.12) with structured query language (SQL). Categorical variables are presented as medians with interquartile ranges and compared by the chi-square test. Continuous variables are reported as median (interquartile range) and compared by the Mann\u0026ndash;Whitney U test. All statistical analyses were performed in R (v4.4.3).\u003c/p\u003e\u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003e3.1 Baseline Characteristics\u003c/h2\u003e\u003cp\u003eAfter applying inclusion and exclusion criteria, 1,890 patients with GIB-AKI were included in the final cohort (684 females, 36.2%; 1,206 males, 63.8%). By day 28 of post-ICU admission, 452 had died, yielding an overall 28-day mortality rate of 23.9%. The patient selection flowchart is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The dataset was randomly divided into a training set (70%, n\u0026thinsp;=\u0026thinsp;1,323) for model construction and a validation set (30%, n\u0026thinsp;=\u0026thinsp;567) for model evaluation. As summarized in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, there were no statistically significant differences in baseline characteristics between the two cohorts (all p\u0026thinsp;\u0026gt;\u0026thinsp;0.05), indicating good comparability.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\u003ch2\u003e3.2 Variable Selection\u003c/h2\u003e\u003cp\u003eTo identify key variables associated with 28-day mortality, we employed a two-step feature selection strategy comprising LASSO regression followed by the Boruta algorithm, both applied to the training set. First, LASSO regression with 10-fold cross-validation identified 26 variables associated with 28-day mortality. The optimal regularization parameter was λ\u0026thinsp;=\u0026thinsp;0.01462515 (log λ = \u0026minus;\u0026thinsp;4.225013) (Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA,B). These variables are listed with their regression coefficients in Supplementary Table\u0026nbsp;1. Next, the Boruta algorithm was utilized to further analyze the 26 variables, assessing their importance. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC, 19 variables, highlighted in green, were retained as significant predictors of 28-day mortality. These 19 variables were incorporated into subsequent model construction and analysis.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003e3.3 Model Development and Performance Evaluation\u003c/h2\u003e\u003cp\u003eThis study established five ML models to predict 28-day all-cause mortality in patients with GIB-AKI. In the training set, the AUC values for Logistic Regression, XGBoost, Decision Tree (DT), Support Vector Machine (SVM), and LightGBM models were 0.831, 0.882, 0.754, 0.844, and 0.848, respectively. The XGBoost model achieved the highest AUC (0.882, 95% CI: 0.862\u0026ndash;0.903), as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. In the validation set, the AUC values for Logistic Regression, XGBoost, Decision Tree (DT), Support Vector Machine (SVM), and LightGBM models were 0.785, 0.803, 0.700, 0.784, and 0.768, respectively. The XGBoost model demonstrated superior performance in the validation set with an AUC of 0.803 (95% CI: 0.759\u0026ndash;0.846), as illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. Comparative performance metrics, including F₁-score, accuracy, precision, and recall, are summarized in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. XGBoost outperformed all other models across these key measures.\u003c/p\u003e\u003cp\u003eCalibration curves (Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e) d revealed that XGBoost predictions closely matched observed mortality rates in both training and validation sets, indicating a model fit exceeding that of the other four models. DCA analysis (Figs.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e) further confirmed that XGBoost yielded the greatest net clinical benefit over a wide range of threshold probabilities in predicting 28-day mortality compared to other models.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003e3.4 Interpretability Analysis\u003c/h2\u003e\u003cp\u003eGiven its superior performance in both training and validation sets, the XGBoost model was subjected to SHapley Additive exPlanations (SHAP) analysis to elucidate feature contributions. Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA presents the SHAP summary plot of the top ten predictors ranked by mean |SHAP| value: APSIII score, AKI stage, Charlson Comorbidity Index, Age, Alkaline Phosphatase (ALP), Vasoactive drug use, Lactate, Diabetes, Anion gap, and Activated Partial Thromboplastin Time (APTT). In this plot, each point represents a single patient; its position on the x-axis reflects that feature\u0026rsquo;s SHAP value (positive values indicate increased mortality risk, while negative values indicate reduced risk), and color denotes the feature\u0026rsquo;s relative magnitude (yellow\u0026thinsp;=\u0026thinsp;high, purple\u0026thinsp;=\u0026thinsp;low). To illustrate individualized prediction, a SHAP force plot for one representative patient is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eB. Elevated APSIII score, higher AKI stage, increased bilirubin, raised PaCO₂ (reflecting respiratory compromise), elevated ALP, and larger mean corpuscular volume (MCV) are visualized as driving upward the patient\u0026rsquo;s predicted 28-day mortality risk. This level of transparency may aid clinicians in understanding and trusting the model\u0026rsquo;s outputs when making patient-specific decisions.\u003c/p\u003e\u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eIn this study, we developed and validated several ML models to predict 28-day all-cause mortality in patients with GIB-AKI, using data from the MIMIC-IV database. Among the five candidate models, the XGBoost model consistently outperformed traditional logistic regression and other ML models across multiple performance metrics, including discrimination, calibration, and clinical net benefit.\u003c/p\u003e\u003cp\u003eThe observed short-term mortality rate in the GIB-AKI cohort was 23.9%, indicating a poor overall prognosis and underscoring the need for accurate risk stratification tools. By combining LASSO regression and the Boruta algorithm for feature selection, we identified 19 key predictive variables. These variables\u0026mdash;APSIII score, AKI stage, Charlson Comorbidity Index, lactate, alkaline phosphatase (ALP), and PaCO₂\u0026mdash;are closely tied to illness severity and organ dysfunction, reflecting real-world clinical considerations.\u003c/p\u003e\u003cp\u003eAmong these predictors, the APSIII score emerged as the most important contributor. As a comprehensive index of disease severity in ICU patients, a higher APSIII score is strongly correlated with increased mortality risk[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e], consistent with previous studies on prognostic factors for all-cause mortality in critically ill patients[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. This finding reaffirms the central role of global physiological derangement in determining short-term outcomes in critically ill populations. AKI not only reflects acute deterioration of renal function but also often indicates systemic inflammatory response, hypovolemia, and tissue hypoxia. AKI stages also show a strong association with increased mortality risk, supporting its established role as an independent risk factor for ICU mortality[\u003cspan additionalcitationids=\"CR18\" citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. The Charlson Index, a vital tool for measuring chronic comorbidity burden, shows that higher scores significantly increase both long-term and short-term mortality risks[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. In this study, the Charlson Comorbidity Index, a well-validated measure of comorbid burden, was a key predictor. A higher Charlson score reflects increased vulnerability due to chronic diseases such as cardiovascular disorders, diabetes, cirrhosis, and malignancies, all of which are known to compromise physiologic reserve during acute stress and increase the risk of multi-organ failure[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eAdditionally, serum lactate was another major predictor. Lactate is a surrogate marker of tissue hypoperfusion and cellular metabolic stress and has been repeatedly linked to adverse outcomes in critically ill patients[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. Its strong contribution to mortality prediction in our model aligns with prior evidence and emphasizes the importance of early identification of circulatory failure[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. PaCO₂ was also among the top predictors. Elevated PaCO₂ may reflect hypoventilation, respiratory failure, or impaired compensatory mechanisms in the context of metabolic acidosis[\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. The positive association between PaCO₂ and mortality risk in our model highlights the prognostic importance of respiratory assessment in GIB-AKI patients.\u003c/p\u003e\u003cp\u003eTraditional GIB risk scoring systems such as the Glasgow-Blatchford Score (GBS), AIMS65, and Rockall Score have proven useful for initial triage and early endoscopy decisions but are inadequate for capturing the complex risk profile of GIB patients complicated by AKI[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. Our study introduces several key innovations: (1) Unlike general-purpose scores, our model is tailored specifically to GIB-AKI patients, resulting in improved precision and clinical relevance; 2) By leveraging a broader set of clinical variables and advanced ML techniques, our model captures complex, nonlinear relationships that are overlooked by traditional scoring methods; 3) The use of SHAP enhances transparency, allowing clinicians to understand how individual variables contribute to each prediction\u0026mdash;an essential feature for clinical adoption.\u003c/p\u003e\u003cp\u003eOur model exhibits strong clinical applicability. Early identification of high-risk GIB-AKI patients upon ICU admission can guide resource allocation, monitoring intensity, and individualized therapy. For instance, patients with high APSIII scores and elevated lactate may benefit from early organ support. Similarly, elevated PaCO₂ should prompt timely respiratory assessment and intervention. Identifying high-risk patients could facilitate early initiation of CRRT, aggressive resuscitation, and enhanced surveillance to mitigate clinical deterioration.\u003c/p\u003e\u003cp\u003eNevertheless, several limitations should be acknowledged. First, this is a single-center study based on the MIMIC-IV database, which may limit generalizability due to center-specific practices and patient populations. External validation in multicenter cohorts is essential. Second, although multiple imputation was employed to address missing data, residual bias from imputed variables cannot be fully excluded. Third, while SHAP enhances interpretability, it reflects statistical associations rather than causal relationships; thus, findings should be interpreted with clinical judgment.\u003c/p\u003e\u003cp\u003eTo enhance model robustness and applicability, future work should pursue 1) external validation across diverse healthcare systems; 2) incorporation of novel biomarkers, such as inflammatory cytokines or metabolomic profiles, alongside clinical variables; and 3) development of dynamic, time-updated prediction frameworks that integrate serial measurements to capture evolving patient trajectories. Ultimately, integrating such tools into electronic health records could enable real-time risk assessment, guiding precision interventions that improve outcomes for GIB-AKI patients.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eWe conducted a retrospective cohort study using data from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database, including 1,890 adult GIB-AKI patients. We applied Least Absolute Shrinkage and Selection Operator (LASSO) regression and the Boruta algorithm to identify 19 key predictors of 28-day mortality. Among the five machine learning (ML) algorithms, XGBoost demonstrated the best discriminative performance. Calibration and decision curve analyses confirmed its superior clinical utility. Furthermore, SHapley Additive exPlanations (SHAP) enhanced model interpretability by illustrating key variables influencing mortality. This study introduces a robust, interpretable ML model for early mortality prediction in GIB-AKI, offering a valuable tool for critical care management and precision medicine practice.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n \u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eBaseline patient characteristics of GIB-AKI patients.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eVariables\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eTraining set(n\u0026thinsp;=\u0026thinsp;1323)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eValidation set(n\u0026thinsp;=\u0026thinsp;567)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eP value\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGeneral information\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge (years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e68.000 [56.000, 79.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e67.000 [56.000, 79.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.789\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGender,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFemale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e475 (35.90)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e209 (36.86)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.730\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e848 (64.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e358 (63.14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRace,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWhite\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e882 (66.67)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e380 (67.02)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.554\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBlack\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e126 (9.52)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e65 (11.46)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAsian\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e36 (2.72)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e15 (2.65)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHispanic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e65 (4.91)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e21 (3.70)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eOther\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e214 (16.18)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e86 (15.17)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWeight(kg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e81.400 [68.750, 95.450]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e80.400 [68.000, 95.700]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.785\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLos ICU(day)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3.260 [1.955, 6.625]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3.060 [1.930, 5.820]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.292\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eArea of bleeding,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLower\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e266 (20.11)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e109 (19.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.787\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUpper\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e830 (62.74)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e354 (62.43)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUnknown\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e227 (17.16)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e104 (18.34)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAKI stage,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e299 (22.60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e124 (21.87)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.696\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eII\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e581 (43.92)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e261 (46.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIII\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e443 (33.48)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e182 (32.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVital signs\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHR (bpm)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e92.000 [79.000, 106.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e94.000 [80.000, 108.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.155\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSBP (mmHg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e118.000 [103.000, 137.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e121.000 [105.000, 138.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.142\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDBP (mmHg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e65.000 [55.000, 77.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e67.000 [56.000, 78.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.268\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMBP (mmHg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e79.000 [69.000, 92.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e82.000 [70.000, 93.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.208\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRR (bpm)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e20.000 [16.000, 24.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e19.000 [16.000, 23.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.132\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTemperaturee(℃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e36.720 [36.440, 37.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e36.720 [36.440, 37.060]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.628\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUrine output(mL)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1200.000 [741.500, 1865.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1215.000 [712.500, 1927.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.995\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSPO2(mmHg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e98.000 [95.000, 100.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e98.000 [95.000, 100.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.395\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003elaboratory test\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eChloride(mmol/l)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e105.000 [100.000, 109.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e104.000 [99.000, 108.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.076\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCalcium(mmol/l)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e8.100 [7.600, 8.700]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e8.100 [7.600, 8.600]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.707\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePotassium(mmol/l)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4.200 [3.800, 4.700]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4.200 [3.800, 4.800]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.857\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSodium(mmol/l)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e139.000 [135.000, 142.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e139.000 [134.500, 142.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.113\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGlucose (mg/dl)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e128.000 [106.000, 169.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e131.000 [106.000, 179.500]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.305\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCreatinine (mg/dL)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.100 [0.800, 1.800]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.200 [0.800, 1.900]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.204\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAniongap(mmol/l)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14.000 [11.000, 18.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14.000 [12.000, 18.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.297\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBUN(mg/dL)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e30.000 [18.500, 49.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e31.000 [18.500, 54.500]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.289\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eALT (IU/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e26.000 [15.000, 51.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e26.000 [15.000, 54.500]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.628\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eALP (IU/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e84.000 [59.000, 125.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e81.000 [57.000, 124.500]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.565\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAST (IU/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e45.000 [25.000, 99.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e43.000 [24.000, 101.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.717\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBilirubin(mg/dL)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.100 [0.500, 3.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.100 [0.500, 2.900]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.594\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWBC (\u0026times;109/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e10.700 [7.100, 15.800]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e11.000 [7.550, 15.800]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.600\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRBC (\u0026times;109/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3.020 [2.550, 3.585]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3.000 [2.560, 3.620]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.815\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePLT (\u0026times;109/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e155.000 [99.500, 238.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e170.000 [105.500, 238.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.291\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHB(g/dL)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e9.000 [7.700, 10.650]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e9.000 [7.700, 10.700]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.847\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMCHC(g/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e32.800 [31.600, 33.900]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e32.900 [31.500, 34.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.785\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRDW(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e16.000 [14.700, 18.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e16.000 [14.650, 18.200]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.717\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMCV(fl)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e92.000 [88.000, 97.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e92.000 [88.000, 97.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.678\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMCH(pg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e30.200 [28.750, 31.900]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e30.200 [28.500, 32.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.706\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHCT(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e27.500 [23.600, 32.400]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e27.700 [23.400, 32.200]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.923\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePT(s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e15.500 [13.400, 20.300]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e15.500 [13.200, 19.600]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.313\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAPTT(s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e31.600 [27.600, 39.050]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e30.800 [26.800, 39.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.182\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eINR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.400 [1.200, 1.900]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.400 [1.200, 1.800]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.374\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLactate (mmol/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.900 [1.300, 3.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2.000 [1.300, 3.150]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.595\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePO2(mmHg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e74.000 [43.000, 136.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e70.000 [42.500, 137.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.578\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePCO2(mmHg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e40.000 [34.000, 47.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e40.000 [35.000, 47.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.842\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e7.360 [7.280, 7.410]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e7.350 [7.270, 7.410]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.120\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBE(mmol/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e-2.000 [-6.000, 0.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e-3.000 [-7.000, 0.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.174\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eComorbidity disease\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMyocardial infarct,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1087 (82.16)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e450 (79.37)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.172\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e236 (17.84)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e117 (20.63)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMalignant cancer,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1131 (85.49)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e478 (84.30)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.554\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e192 (14.51)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e89 (15.70)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCharlson comorbidity index\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e6.000 [4.000, 8.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e6.000 [4.000, 8.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.827\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSevere liver disease,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e913 (69.01)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e391 (68.96)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e410 (30.99)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e176 (31.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRenal disease,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e973 (73.54)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e424 (74.78)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.615\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e350 (26.46)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e143 (25.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCongestive heart failure,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e899 (67.95)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e392 (69.14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.651\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e424 (32.05)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e175 (30.86)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePeripheral vascular disease,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1158 (87.53)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e496 (87.48)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e165 (12.47)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e71 (12.52)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCerebrovascular disease,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1178 (89.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e504 (88.89)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.987\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e145 (10.96)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e63 (11.11)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eChronic pulmonary disease,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e997 (75.36)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e433 (76.37)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.682\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e326 (24.64)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e134 (23.63)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePeptic ulcer disease,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e888 (67.12)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e381 (67.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e435 (32.88)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e186 (32.80)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDiabetes,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e917 (69.31)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e384 (67.72)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.530\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e406 (30.69)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e183 (32.28)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTreatment and severity scores\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVasoactive,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e800 (60.47)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e341 (60.14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.935\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e523 (39.53)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e226 (39.86)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSOFA (score)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e5.00 [2.00, 8.00]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e5.00 [2.00, 8.00]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.938\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGCS (score)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14.00 [12.00, 15.00]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14.0 [13.0, 15.0]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.154\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAPSIII (score)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e50.000 [40.000, 66.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e52.000 [41.000, 67.000]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.303\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVentilator,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e235 (17.76)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e97 (17.11)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.782\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1088 (82.24)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e470 (82.89)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEndoscopy therpay,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1221 (92.29)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e530 (93.47)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.419\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e102 (7.71)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e37 (6.53)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCRRT,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1198 (90.55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e507 (89.42)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.499\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e125 (9.45)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e60 (10.58)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTransfusion of red cells,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1289 (97.43)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e549 (96.83)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.560\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e34 (2.57)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e18 (3.17)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eLos ICU, length of ICU stay;HR, heart rate; SBP, systolic blood pressure; DBP, diastolic blood pressure; MAP, mean arterial pressure; RR, respiratory rate; BUN, blood urea nitrogen;ALT, alanine aminotransferase;AST, aspartate aminotransferase; ALP,alkaline phosphatase;RBC, red blood cell; WBC, white blood cell; PLT, platelet; RDW, erythrocyte distribution width; HB, hemoglobin; HCT, hematocrit;PT, prothrombin time; INR, international normalized ratio;BE,Base Excess;APTT, Activated Partial Thromboplastin Time;GCS, Glasgow coma scale; SOFA, sequential organ failure assessment; APSS III, Acute Physiology Score System III.CRRT, continuous renal replacement therapy; AKI, acute kidney injury.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n \u003ctable id=\"Tab2\" border=\"1\" class=\"fr-table-selection-hover\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eEvaluation of five models performance in the training set and validation set.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eModel\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSensitivity\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSpecificity\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eF1\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTraining set\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003elightGBM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.810\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.769\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.823\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.659\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLogistic Regression\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.752\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.832\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.727\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.616\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSVM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.822\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.794\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.831\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.681\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDecision Tree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.766\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.759\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.769\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.608\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.835\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.839\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.834\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.709\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eValidation set\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003elightGBM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.771\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.654\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.807\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.578\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLogistic Regression\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.732\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.787\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.715\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.585\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSVM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.771\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.706\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.791\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.596\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDecision Tree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.757\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.706\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.773\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.582\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.790\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.765\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.798\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.636\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eLos ICU, length of ICU stay;HR, heart rate; SBP, systolic blood pressure; DBP, diastolic blood pressure; MAP, mean arterial pressure; RR, respiratory rate; BUN, blood urea nitrogen;ALT, alanine aminotransferase;AST, aspartate aminotransferase; ALP,alkaline phosphatase;RBC, red blood cell; WBC, white blood cell; PLT, platelet; RDW, erythrocyte distribution width; HB, hemoglobin; HCT, hematocrit;PT, prothrombin time; INR, international normalized ratio;BE,Base Excess;APTT, Activated Partial Thromboplastin Time;GCS, Glasgow coma scale; SOFA, sequential organ failure assessment; APSS III, Acute Physiology Score System III.CRRT, continuous renal replacement therapy; AKI, acute kidney injury.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets generated and analysed during the current study are available in the MIMIC-IV (v. 3.1) database, which contains deidentified data. The database complies with the ethical guidelines outlined in the Declaration of Helsinki. Therefore, additional approval from the institutional review board was not required. Access to the database was granted to the research team under user ID: 70784184.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp skip=\"true\"\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data presented in the current study are available in the MIMIC-IV database (https://physionet.org/content/mimiciv/).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eNo funding support.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors and Affiliations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDepartment of Gastrointestinal Surgery, Qilu Hospital of Shandong University Dezhou Hospital, 1751 Xinhu Street, Dezhou 253000, China\u003c/p\u003e\n\u003cp\u003eXiangyu Zhang,Yanpeng Hu,Xingye Zhu,Chan Yu,Cuicui Liu,Jian Xue,Yingfeng Su,Baoqing Ma.\u003c/p\u003e\n\u003cp\u003eXiangyu Zhang and Yanpeng Hu have contributed equally to this work and share the first authorship.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contribution\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConceptualization: Xiangyu Zhang, Baoqing Ma.\u003c/p\u003e\n\u003cp\u003eData curation: Xiangyu Zhang, Cuicui Liu, Yanpeng Hu.\u003c/p\u003e\n\u003cp\u003eFormal analysis: Yanpeng Hu, Xingye Zhu, Jian Xue.\u003c/p\u003e\n\u003cp\u003eWriting\u0026mdash;original draft: Xiangyu Zhang, Baoqing Ma, Yingfeng Su.\u003c/p\u003e\n\u003cp\u003eWriting\u0026mdash;review and editing: Xingye Zhu, Chan Yu,Baoqing Ma.\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eCorresponding author\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eCorrespondence to\u0026nbsp;Baoqing Ma.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe are grateful to the MIMIV-IV participants and staff. We appreciate all the reviewers who participated in the review.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eLong B, Gottlieb M. Emergency medicine updates: Upper gastrointestinal bleeding. Am J Emerg Med. 2024;81:116-123. \u003c/li\u003e\n\u003cli\u003eHong MJ,Lee SY,Kim JH, et al. Rebleeding after initial endoscopic hemostasis in peptic ulcer disease. J Korean Med Sci. 2014;29 (10):1411-5. \u003c/li\u003e\n\u003cli\u003eMorarasu BC, Sorodoc V, Haisan A, et al. Age, blood tests and comorbidities and AIMS65 risk scores outperform Glasgow-Blatchford and pre-endoscopic Rockall score in patients with upper gastrointestinal bleeding. World J Clin Cases. 2023;11(19):4513-4530.\u003c/li\u003e\n\u003cli\u003eZhong L,Quan X,Dang P, et al. Clinical characteristics and risk factors of in-hospital gastrointestinal bleeding in patients with acute myocardial infarction. Front Cardiovasc Med. 2022;9:933597. \u003c/li\u003e\n\u003cli\u003eCakmak U, Merhametsiz O, Gok Oguz E, et al. Effects of acute kidney injury on clinical outcomes in patients with upper gastrointestinal bleeding. Ren Fail. 2016;38(2):176-184. \u003c/li\u003e\n\u003cli\u003eZhang H,Wang Z,Tang Y, et al. Prediction of acute kidney injury after cardiac surgery: model development using a Chinese electronic health record dataset. J Transl Med. 2022;20 (1):166. \u003c/li\u003e\n\u003cli\u003eCastellano G,Franzin R,Sallustio F, et al. Complement component C5a induces aberrant epigenetic modifications in renal tubular epithelial cells accelerating senescence by Wnt4/\u0026beta;catenin signaling after ischemia/reperfusion injury. Aging (Albany NY). 2019;11 (13):4382-4406.\u003c/li\u003e\n\u003cli\u003eKwak MS,Cha JM,Han YJ, et al. The Clinical Outcomes of Lower Gastrointestinal Bleeding Are Not Better than Those of Upper Gastrointestinal Bleeding. J Korean Med Sci. 2016;31 (10):1611-6.\u003c/li\u003e\n\u003cli\u003eWan X,Xie X,Gendoo Y, et al. Ulinastatin administration is associated with a lower incidence of acute kidney injury after cardiac surgery: a propensity score matched study. Crit Care. 2016;20:42. \u003c/li\u003e\n\u003cli\u003eTantai XX,Liu N,Yang LB, et al. Prognostic value of risk scoring systems for cirrhotic patients with variceal bleeding. World J Gastroenterol. 2019;25 (45):6668-6680.\u003c/li\u003e\n\u003cli\u003eAssaf D,Gutman Y,Neuman Y, et al. Utilization of machine-learning models to accurately predict the risk for critical COVID-19. Intern Emerg Med. 2020;15 (8):1435-1443. \u003c/li\u003e\n\u003cli\u003eExplainable Machine-Learning Model for Prediction of In-Hospital Mortality in Septic Patients Requiring Intensive Care Unit Readmission. Infect Dis Ther. 2022;11 (4):1695-1713. \u003c/li\u003e\n\u003cli\u003eMeersch M,Schmidt C,Hoffmeier A, et al. Prevention of cardiac surgery-associated AKI by implementing the KDIGO guidelines in high risk patients identified by biomarkers: the PrevAKI randomized controlled trial. Intensive Care Med. 2017;43 (11):1551-1561. \u003c/li\u003e\n\u003cli\u003eXu Y,Han D,Huang T, et al. Predicting ICU Mortality in Rheumatic Heart Disease: Comparison of XGBoost and Logistic Regression. Front Cardiovasc Med. 2022;9:847206. \u003c/li\u003e\n\u003cli\u003eLangelier C,Kalantar KL,Moazed F, et al. Integrating host response and unbiased microbe detection for lower respiratory tract infection diagnosis in critically ill adults. Proc Natl Acad Sci U S A. 2018;115 (52):E12353-E12362. \u003c/li\u003e\n\u003cli\u003eXue G,Liang H,Ye J, et al. Development and Validation of a Predictive Scoring System for In-hospital Death in Patients With Intra-Abdominal Infection: A Single-Center 10-Year Retrospective Study. Front Med (Lausanne). 2021;8:741914.\u003c/li\u003e\n\u003cli\u003eHosseini-Moghaddam SM,Luo B,Bota SE, et al. Incidence and Outcomes Associated With Clostridioides difficile Infection in Solid Organ Transplant Recipients. JAMA Netw Open. 2021;4 (12):e2141089. \u003c/li\u003e\n\u003cli\u003eYu WQ,Zhang SY,Fu SQ, et al. Dexamethasone protects the glycocalyx on the kidney microvascular endothelium during severe acute pancreatitis. J Zhejiang Univ Sci B. 2019;20 (4):355-362. \u003c/li\u003e\n\u003cli\u003eHaase-Fielitz A,Altendeitering F,Iwers R, et al. Acute kidney injury may impede results after transcatheter aortic valve implantation. Clin Kidney J. 2021;14 (1):261-268. \u003c/li\u003e\n\u003cli\u003eKallogjeri D,Gaynor SM,Piccirillo ML, et al. Comparison of comorbidity collection methods. J Am Coll Surg. 2014;219 (2):245-55. \u003c/li\u003e\n\u003cli\u003eYu Q,Wang Y,Huang S, et al. Multicenter cohort study demonstrates more consolidation in upper lungs on initial CT increases the risk of adverse clinical outcome in COVID-19 patients. Theranostics. 2020;10 (12):5641-5648. \u003c/li\u003e\n\u003cli\u003eSchierenbeck F,Nijsten MW,Franco-Cereceda A, et al. Introducing intravascular microdialysis for continuous lactate monitoring in patients undergoing cardiac surgery: a prospective observational study. Crit Care. 2014;18 (2):R56. \u003c/li\u003e\n\u003cli\u003eGao F,Huang XL,Cai MX, et al. Prognostic value of serum lactate kinetics in critically ill patients with cirrhosis and acute-on-chronic liver failure: a multicenter study. Aging (Albany NY). 2019;11 (13):4446-4462. \u003c/li\u003e\n\u003cli\u003eDaga MK,Rohatgi I,Mishra R, et al. Lactate enhanced-quick Sequential Organ Failure Assessment 2 (LqSOFA2): A new score for bedside prognostication of patients with sepsis. Indian J Med Res. 2021;154 (4):607-614. \u003c/li\u003e\n\u003cli\u003eCaruso V,Besch G,Nguyen M, et al. Treatment of Hyperlactatemia in Acute Circulatory Failure Based on CO 2 -O 2 -Derived Indices: Study Protocol for a Prospective, Multicentric, Single, Blind, Randomized, Superiority Study (The LACTEL Study). Front Cardiovasc Med. 2022;9:898406. \u003c/li\u003e\n\u003cli\u003eLee SH,Park S,Lee JW, et al. The Anion Gap is a Predictive Clinical Marker for Death in Patients with Acute Pesticide Intoxication. J Korean Med Sci. 2016;31 (7):1150-9. \u003c/li\u003e\n\u003cli\u003eRobin E,Futier E,Pires O, et al. Central venous-to-arterial carbon dioxide difference as a prognostic tool in high-risk surgical patients. Crit Care. 2015;19:227. \u003c/li\u003e\n\u003cli\u003eShi H, Shen Y, Li L. Early prediction of acute kidney injury in patients with gastrointestinal bleeding admitted to the intensive care unit based on extreme gradient boosting. Front Med (Lausanne). 2023;10:1221602.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-gastroenterology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmge","sideBox":"Learn more about [BMC Gastroenterology](http://bmcgastroenterol.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmge/default.aspx","title":"BMC Gastroenterology","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Gastrointestinal bleeding, Acute kidney injury, Machine learning","lastPublishedDoi":"10.21203/rs.3.rs-7339661/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7339661/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground: \u003c/strong\u003eGastrointestinal bleeding (GIB) is a common life-threatening condition in the digestive system that is frequently complicated by acute kidney injury (AKI), substantially increasing mortality and healthcare burden. To date, no precise tool exists for early prediction of short-term outcomes in patients with concurrent GIB and AKI (GIB-AKI).We conducted a retrospective cohort study aims to develop and validate machine learning (ML) models for predicting 28-day mortality .\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eThis retrospective cohort study was based on the MIMIC-IV database, including patients with first ICU admission who met criteria for GIB and AKI .From 64 clinical variables, we applied Least Absolute Shrinkage and Selection Operator (LASSO) regression and the Boruta algorithm to identify the key predictors of 28-day mortality. Five ML algorithms were employed to construct the prediction model. Model performance was evaluated based on identification, calibration, and clinical application.SHapley Additive exPlanations (SHAP) were used to interpret key variables influencing mortality in the best model.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eA total of 1,890 adult GIB-AKI patients were included in this study.\u003c/p\u003e\n\u003cp\u003eFive machine learning (ML) algorithms—Light Gradient Boosting Machine (LightGBM), Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), Logistic Regression, and Decision Tree—were developed and compared. Among all models, XGBoost demonstrated the best discriminative performance, achieving an area under the receiver operating characteristic curve (AUC) of 0.8027 in the validation set. Calibration and decision curve analyses confirmed its superior clinical utility. Furthermore, SHapley Additive exPlanations (SHAP)enhanced model interpretability by illustrating key variables influencing mortality.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion: \u003c/strong\u003eThis study introduces a robust, interpretable ML model for early mortality prediction in GIB-AKI, offering a valuable tool for critical care management and precision medicine practice.\u003c/p\u003e","manuscriptTitle":"Machine Learning Models for Predicting 28-Day Mortality in Gastrointestinal Bleeding with Acute Kidney Injury: A MIMIC-IV-Based Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-01 10:07:41","doi":"10.21203/rs.3.rs-7339661/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewersInvited","content":"","date":"2025-09-19T11:29:34+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-09-01T06:15:54+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-08-25T08:44:49+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-08-23T14:24:59+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Gastroenterology","date":"2025-08-23T14:21:49+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-gastroenterology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmge","sideBox":"Learn more about [BMC Gastroenterology](http://bmcgastroenterol.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmge/default.aspx","title":"BMC Gastroenterology","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"78b78ae5-031f-4449-9909-c62414c00270","owner":[],"postedDate":"October 1st, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-10-01T10:07:41+00:00","versionOfRecord":[],"versionCreatedAt":"2025-10-01 10:07:41","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7339661","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7339661","identity":"rs-7339661","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}