Machine learning models and restricted cubic spline were employed to analyze and predict postoperative ischemic stroke in type A aortic dissection patients

Machine learning models and restricted cubic spline were employed to analyze and predict postoperative ischemic stroke in type A aortic dissection patients | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Machine learning models and restricted cubic spline were employed to analyze and predict postoperative ischemic stroke in type A aortic dissection patients WenJian Ma, Siji Chen, Yang Zhao, Shuanglei Zhao, Qianxian Li, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7277000/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 10 Dec, 2025 Read the published version in BMC Cardiovascular Disorders → Version 1 posted 14 You are reading this latest preprint version Abstract BACKGROUD : Ischemic stroke remains a devastating postoperative complication in Type A aortic dissection (TAAD) patients, contributing significantly to elevated mortality rates. Identifying reliable predictors for ischemic stroke risk is crucial for implementing timely clinical interventions. This study endeavors to develop and validate a machine learning-based predictive model for ischemic stroke risk stratification in TAAD patients undergoing surgical treatment. Methods : This retrospective cohort study analyzed 430 TAAD patients who underwent total aortic arch replacement with frozen elephant trunk implantation at Beijing Anzhen Hospital (2015-2021). The cohort was randomly partitioned into training (70%, n=301) and validation (30%, n=129) sets. Feature selection was performed using Boruta algorithm, with variables demonstrating P<0.1 in univariate analysis subsequently incorporated into multivariate logistic regression. Ten machine learning models were evaluated through receiver operating characteristic (ROC) curves, decision curve analysis (DCA), and calibration plots. Model interpretability was enhanced via Shapley Additive Explanations (SHAP), while restricted cubic splines (RCS) elucidated potential non-linear/liner relationships between predictors and result. Results: The GBM model demonstrated superior predictive performance compared to all other models, achieving an area under the curve (AUC) of 0.804 in the validation cohort. SHAP analysis identified the following key predictors of postoperative ischemic stroke: age, history of cerebrovascular disease, cardiopulmonary bypass time(CPBT), intraoperative blood loss volume(IBLV), and preoperative systolic blood pressure(SBP).Furthermore,RCS were independently constructed for each continuous variable to explore variable-outcome relationships. Conclusion: The Gradient Boosting Machine (GBM) model demonstrates the best predictive capacity for postoperative ischemic stroke in TAAD patients, offering clinicians a clinically actionable tool for early postoperative risk stratification and personalized therapeutic optimization. Machine learning model Risk factors Ischemic stroke Surgical treatment Type A aortic dissection Figures Figure 1 Figure 2 Figure 3 Figure 5 Figure 6 Figure 7 Figure 8 Introduction Type A aortic dissection (TAAD), a life-threatening cardiovascular emergency disease characterized by end-organ mal-perfusion, exhibits marked time-dependent progression. Studies demonstrate an hourly mortality rate of 1-2% following symptom onset, with emergency surgical mortality decreasing from 5.8% (baseline) to 4.4% post-intervention within 48 hours [1] . Despite advancements in modern aortic surgery—including Sun's procedure [2] that integrates total arch replacement with specialized stent-graft deployment in the descending aorta, central arterial repair techniques, and neuroprotective strategies such as selective cerebral perfusion combined with moderate hypothermic circulatory arrest—the incidence of postoperative neurological complication syndrome (NCs) remains persistently high at 17%-48% [3] .Postoperative NCs following emergency TAAD repair are associated with prolonged intensive care/hospital stays (ischemic stroke: 23 ± 16 days vs no ischemic stroke: 17 ± 18 days, P = 0.021) and morbidity [4, 5] .Ischemic stroke, the most clinically significant NCs, demonstrated a postoperative incidence of 24.8% in our prior study [6] .The development of early, precise predictive models for in-hospital neurological complications carries urgent clinical significance for optimizing decision-making and improving outcomes in this critical population. While multiple risk factors associated with postoperative ischemic stroke have been identified in existing studies, there remains an urgent need for reliable, data-driven predictive models to systematically identify in-hospital ischemic stroke occurrence following surgical interventions. Furthermore, validated prediction tools specifically targeting ischemic stroke complications in patients undergoing Sun's procedure (total arch replacement with frozen elephant trunk implantation) are critically lacking in both Chinese and international clinical practice. The development of a precise risk stratification model is therefore imperative to advance preoperative assessment, optimize preventive strategies, and guide therapeutic decision-making, ultimately aiming to reduce postoperative ischemic stroke incidence and improve survival rates through evidence-based interventions. Machine learning (ML), a specialized subset of artificial intelligence (AI), enables automated extraction of clinically actionable insights for critical tasks including risk stratification, diagnostic classification, and survival prediction. ML algorithms have thus emerged as indispensable tools in biomedical research [7] , demonstrating capabilities to identify latent patterns within complex datasets and generate predictive outputs through advanced feature engineering [8] . Comparative analyses reveal ML's superior performance metrics over conventional statistical methods, with successful clinical implementations [9] and real-time treatment optimization [10] . Materials and Methods 2.1Data Source This retrospective study enrolled patients diagnosed with TAAD who underwent total aortic arch replacement with frozen elephant trunk implantation (Sun's procedure) at the Cardiac Surgery Center of Beijing Anzhen Hospital, Capital Medical University, between 2015 and 2020. A total of 430 consecutive cases with complete perioperative data were included. The primary endpoint was postoperative ischemic stroke occurrence. Diagnosis adhered to the 2014 ESC guidelines [11] , incorporating Stanford classification and confirmatory imaging magnetic resonance angiography (MRA) or computed tomography angiography (CTA). Exclusion criteria comprised:1.Non-surgically managed TAAD 2.Concurrent malignancies with limited life expectancy 3.Acute myocardial infarction secondary to severe myocardial mal-perfusion 4.Incomplete medical records. Postoperative ischemic stroke diagnosis followed internationally recognized ISCHEMIC STROKE guideline criteria [12] .The cohort was stratified into training (70%, n = 301) and validation (30%, n = 129) sets using stratified random sampling to balance baseline characteristics and mitigate overfitting risks. Ethical Compliance Approved by the Institutional Review Board of Beijing Anzhen Hospital (Approval No: 2025124X), this study strictly adhered to the ethical principles of the Declaration of Helsinki. 2.2 Features Extraction The extracted variables encompassed demographic characteristics (age, BMI), clinical profiles (hypertension, history of coronary artery disease, diabetes, smoking/alcohol use, history of cerebrovascular disease, renal insufficiency, prior cardiac surgery, NYHA functional class, acute renal dysfunction), surgical parameters (intraoperative blood loss, cardiopulmonary bypass duration, deep hypothermic circulatory arrest time, aortic cross-clamp time exceeding 3 hours), and ultrasonographic findings including dissection involvement of the left subclavian artery, innominate artery, left common carotid artery, along with true/false lumen perfusion patterns. Laboratory biomarkers comprised hepatic function indices (ALT, AST), coagulation markers (D-dimer), myocardial injury indicators (myoglobin), and renal function tests (serum creatinine). Blood samples were collected within 24 hours of admission, with the initial measurement utilized for variables requiring repeated assessments to ensure temporal consistency and minimize intervention-related confounding. 2.3 Surgical Technique Sun's procedure refers to total arch replacement using a four-branched vascular graft combined with specialized stent-graft implantation in the descending aorta, as technically detailed in the Annals of Thoracic Surgery [13] . Briefly, the procedure was performed under moderate hypothermia at 25°C with circulatory arrest. Cardiopulmonary bypass was established via right axillary artery cannulation, incorporating selective antegrade cerebral perfusion. The surgical steps included: stent-graft deployment in the descending aorta and total arch replacement with a four-branched graft. A specific sequence of vascular reconstruction was followed: proximal descending aorta anastomosis first, followed by the left carotid artery, ascending aorta, left subclavian artery, and finally the celiac artery. Early rewarming and reperfusion were initiated after completion of the distal anastomosis to minimize cerebral and coronary ischemic time. 2.4 Model Construction and Validation Figure 1 displays the concise flowchart of predictive model construction and validation. Spearman correlation analysis was employed to investigate the interrelationships among the variables. The correlation heatmap (Fig. 2 ) illustrated the correlation between each factor, like the correlation coefficient between N and T is 0.36, which is less than 0.5, indicating a weak correlation, and the others are also weak correlations. Collinearity arises when two or more predictor variables exhibit strong correlation, thereby complicating the assessment of each variable’s distinct contribution to the outcome. So, we selected the most readily available variables among the collinear variables for further analysis. while Fig. 3 displays the AUC values for all predictors. Feature selection was conducted using the Boruta algorithm to identify potential risk factors within the training dataset. Boruta’s algorithm is a method used to determine the most important features in a dataset. It identifies importance by comparing the Z value of each feature with the Z value of the corresponding “shadow feature”. In the algorithm, all real features are copied and shuffled, and then the Z value of each feature is obtained through the random forest model. Additionally, the Z values of the ‘shadow features’ are generated by randomly shuffling the real features. Feature selection based on the Boruta algorithm. The horizontal axis is the name of each variable, and the vertical axis is the Z value of each variable. The box plot shows the Z value of each variable during model calculation. The green boxes represent important variables, and the red boxes represent unimportant variables(Fig. 4 ). The selected significant variables were incorporated into ten distinct machine learning algorithms for model construction, including Logistic Regression (LR), K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Gradient Boosting Machine (GBM), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Artificial Neural Network (NNET), Adaptive Boosting (AdaBoost), Light Gradient Boosting Machine (LightGBM), and Categorical Boosting (CatBoost). Model performance was evaluated by selecting the algorithm achieving the highest AUC, with supplementary assessments of discriminatory power conducted through sensitivity, specificity, accuracy, and F1-score metrics. DCA curve quantified clinical utility by estimating net benefit across probability thresholds. The top-performing model underwent interpretability analysis using SHAP. Following parameter optimization on the training cohort, all model hyperparameters were fixed and externally validated on an independent validation cohort to ensure generalizability and mitigate overfitting risks. The interpretability of machine learning models remains a significant challenge in clinical applications. To elucidate how individual features contribute to predictions in our best-performing black-box model, we employed SHAP values—a game theory-based approach that quantifies each feature's impact on model outputs by treating features as collaborative players. SHAP fairly attributes predictive contributions to each variable, enabling both global interpretation through mean absolute SHAP value ranking and local explanation via individual prediction decomposition. Feature importance was determined by calculating the mean absolute SHAP value across all observations. Additionally, we visualized force plots and summary plots to delineate the directional effects and magnitude ranges of key predictors, with violin plots further characterizing non-linear relationships between continuous variables and ischemic stroke risk . 2.5 Statistical Analysis The study population exhibited complete data integrity with no missing values, obviating the need for imputation. All statistical analyses were conducted using R software version 4.4.2. Normally distributed continuous variables are presented as mean ± standard deviation (SD), with between-group comparisons performed using Student's t-test. Non-normally distributed variables are expressed as median (interquartile range, IQR) and analyzed via the Mann-Whitney U test. Categorical variables are reported as counts (percentages), with group differences assessed using Pearson's chi-square test or Fisher's exact test as appropriate for cell frequencies. Result 3.1:Baseline characteristics Following rigorous screening, this study enrolled 430 patients, with 302 allocated to the training set and 128 to the testing set. The postoperative ischemic stroke incidence was 17% in both cohorts. The overall cohort demonstrated a median age of 48.0 years (40.0–56.0), with a male predominance (81% female vs. 19% male). Table 1 details baseline characteristics including demographic parameters, vital signs, and laboratory indices. Comparative analysis revealed statistically significant differences (P < 0.05) between ischemic stroke and non-ischemic stroke groups: Preoperative Factors: Ischemic stroke patients were older (median age 51 vs. 48 years, P = 0.003) with higher systolic blood pressure (140 vs130 mmHg, P = 0.004) and greater prevalence of cerebrovascular disease history (9.4% vs. 2.0%, P = 0.017).Biochemical Profile: Elevated myoglobin levels (median 32.1vs. 43.8 ng/mL, P = 0.012) and impaired hepatic function (AST: 22 vs. 27U/L, P = 0.021) were observed in ischemic stroke cases. Intraoperative Metrics: Prolonged cardiopulmonary bypass time (CPBT: 201vs224, P < 0.001) and increased intraoperative blood loss (IBV: 1200 vs. 1500mL, P < 0.001). Table 1 Patient Demographics and Baseline Characteristics in the Training Cohort Characteristic Overall N = 302 No stroke N = 249 stroke N = 53 p-value1 sex, n (%) 0.6 male 56 (19%) 45 (18%) 11 (21%) female 246 (81%) 204 (82%) 42 (79%) hypertension, n (%) 246 (81%) 203 (82%) 43 (81%) > 0.9 diabete, n (%) 19 (6.3%) 19 (7.6%) 0 (0%) 0.031 CBDH, n (%) 10 (3.3%) 5 (2.0%) 5 (9.4%) 0.017 Renal insufficiency history, n (%) 6 (2.0%) 5 (2.0%) 1 (1.9%) > 0.9 CHD history, n (%) 21 (7.0%) 15 (6.0%) 6 (11%) 0.2 smoke, n (%) 144 (48%) 122 (49%) 22 (42%) 0.3 drink, n (%) 76 (25%) 64 (26%) 12 (23%) 0.6 CSH, n (%) 15 (5.0%) 11 (4.4%) 4 (7.5%) 0.3 NYHA class, n (%) 0.3 0 178 (59%) 141 (57%) 37 (70%) I 0 (0%) 0 (0%) 0 (0%) II 64 (21%) 55 (22%) 9 (17%) III 34 (11%) 31 (12%) 3 (5.7%) IV 26 (8.6%) 22 (8.8%) 4 (7.5%) AKI, n (%) 25 (8.3%) 19 (7.6%) 6 (11%) 0.4 Brachiocephalic artery, n (%) 0.4 No 143 (47%) 122 (49%) 21 (40%) True 153 (51%) 122 (49%) 31 (58%) False 6 (2.0%) 5 (2.0%) 1 (1.9%) LCAA, n (%) 0.9 No 189 (63%) 154 (62%) 35 (66%) True 110 (36%) 92 (37%) 18 (34%) False 3 (1.0%) 3 (1.2%) 0 (0%) LSA, n (%) 0.5 No 173 (57%) 139 (56%) 34 (64%) True 123 (41%) 105 (42%) 18 (34%) False 6 (2.0%) 5 (2.0%) 1 (1.9%) Limb ischemia, n (%) 37 (12%) 28 (11%) 9 (17%) 0.2 ACCT, n (%) 0.091 3h 17 (5.6%) 11 (4.4%) 6 (11%) AMI, n (%) 27 (8.9%) 20 (8.0%) 7 (13%) 0.3 Age(years), Median (Q1, Q3) 48.00 (40.00, 56.00) 48.00 (37.00, 55.00) 51.00 (43.00, 62.00) 0.003 BMI(kg/m²), Median (Q1, Q3) 26.12 (23.88, 28.41) 26.00 (23.66, 28.40) 26.12 (24.22, 29.35) 0.3 Rate(bpm), Median (Q1, Q3) 84.00 (76.00, 93.00) 84.00 (76.00, 93.00) 82.00 (75.00, 96.00) 0.7 SBP(mmhg), Median (Q1, Q3) 130.50(118.00, 146.00) 130.00(117.00, 144.00) 140.00(127.00, 159.00) 0.004 DBP(mmhg), Median (Q1, Q3) 69.00 (58.00, 78.00) 69.00 (58.00, 78.00) 68.00 (54.00, 77.00) 0.6 LVEDD(mm), Median (Q1, Q3) 50.00 (45.00, 54.00) 50.00 (45.00, 54.00) 49.00 (44.00, 54.00) 0.9 LVEF(%), Median (Q1, Q3) 62.00 (60.00, 66.00) 62.00 (60.00, 66.00) 62.00 (58.00, 66.00) 0.7 DD(ug/mL), Median (Q1, Q3) 2,197.00(972.00,3,373.00) 2,202.00(930.00,3,309.00) 2,138.00(1,118.00,3,409.00) 0.4 MB(ng/mL), Median (Q1, Q3) 33.35 (20.00, 66.40) 32.10 (19.20, 60.70) 43.80 (27.00, 100.80) 0.012 TNI(ng/mL), Median (Q1, Q3) 0.01 (0.00, 0.06) 0.01 (0.00, 0.04) 0.02 (0.01, 0.12) 0.090 ALT(U/L), Median (Q1, Q3) 23.00 (16.00, 33.00) 22.00 (16.00, 33.00) 25.00 (17.00, 38.00) 0.2 AST(U/L), Median (Q1, Q3) 22.50 (18.00, 31.00) 22.00 (18.00, 30.00) 27.00 (20.00, 35.00) 0.013 CR(mg/dL), Median (Q1, Q3) 81.65 (69.30, 100.40) 81.20 (69.20, 98.80) 82.50 (72.00, 104.10) 0.5 CPBT(min), Median (Q1, Q3) 205.00 (180.00, 234.00) 201.00 (179.00, 225.00) 224.00 (198.00, 275.00) < 0.001 SCPT(min), Median (Q1, Q3) 36.00 (29.00, 44.00) 36.00 (29.00, 44.00) 38.00 (30.00, 47.00) 0.3 DHCAT(min), Median (Q1, Q3) 22.00 (18.00, 29.00) 22.00 (18.00, 28.00) 24.00 (18.00, 30.00) 0.2 IBLV(ml), Median (Q1, Q3) 1,300.00 (1,000.00, 1,600.00) 1,200.00 (1,000.00, 1,500.00) 1,500.00 (1,200.00, 2,000.00) < 0.001 1 Pearson's Chi-squared test; Fisher's exact test; Wilcoxon rank sum test 2 Abbreviations :CBDH: cerebrovascular disease history;CHD history:coronary heart disease history;CSH:cardiac surgery history;NYHA class:New York Heart Association (NYHA) Classification;SBP:systolic blood pressure;DBP:diastolic blood pressure;LVEF:Left Ventricular Ejection Fraction;DD:D-Dimer;MB:Myoglobin;TNI:Troponin I;ALT:Alanine Aminotransferase;AST:Aspartate Aminotransferase;LVEDD:Left Ventricular End-Diastolic Diameter;CR:Creatinine;CPBT:Cardiopulmonary Bypass Time;SCPT:Selective cerebral perfusion time;DHCAT:Deep hypothermic circulatory arrest time;IBLV:Intraoperative blood loss volume;AKI:Acute Kidney Injury;LCCA: Left common carotid artery;LSA:Left subclavian artery;ACCT:Aortic cross-clamp time;BA:Brachiocephalic artery Table 2 Multivariate Logistic Regression Analysis for Sepsis in the Training Cohort Characteristic desc No (N = 249) Yes (N = 53) OR (univariable) OR (multivariable) age Mean ± SD 46.8 ± 11.1 52.1 ± 11.8 1.04 (1.01–1.07, p = .003) 1.04 (1.01–1.07, p = .016) CPBT Mean ± SD 205.4 ± 41.9 239.8 ± 63.5 1.01 (1.01–1.02, p 3h 238 (95.6%) 47 (88.7%) < 3h 11 (4.4%) 6 (11.3%) 2.76 (0.97–7.84, p = .056) 1.18 (0.31–4.54, p = .810) IBLV Mean ± SD 1392.8 ± 644.4 1824.5 ± 950.3 1.00 (1.00–1.00, p < .001) 1.00 (1.00–1.00, p = .056) SBP Mean ± SD 131.7 ± 23.4 143.2 ± 27.6 1.02 (1.01–1.03, p = .002) 1.02 (1.01–1.04, p = .007) CHD history No 244 (98%) 48 (90.6%) Yes 5 (2%) 5 (9.4%) 5.08 (1.42–18.24, p = .013) 5.80 (1.09–30.95, p = .040) MB Mean ± SD 94.8 ± 265.2 167.0 ± 529.6 1.00 (1.00–1.00, p = .179) 1.00 (1.00–1.00, p = .320) diabete No 230 (92.4%) 53 (100%) Yes 19 (7.6%) 0 (0%) 0.00 (0.00-Inf, p = .986) 0.00 (0.00-Inf, p = .984) DBP Mean ± SD 68.0 ± 14.0 68.0 ± 15.6 1.00 (0.98–1.02, p = .982) 0.98 (0.96–1.01, p = .245) Abbreviations : ACCT༚Aortic cross-clamp time;CPBT:Cardiopulmonary Bypass Time;IBLV:Intraoperative blood loss volume;SBP:systolic blood pressure;CHD:history:coronary heart disease history;MB:Myoglobin;DBP:diastolic blood pressure Table 3 :Comparison of Testing Cohort Results of the Machine Learning Models Model AUC Accuracy Sensitivity Specificity F1 LR 0.765 0.719 0.773 0.708 0.486 SVM 0.505 0.773 0.273 0.877 0.293 GBM 0.804 0.742 0.864 0.717 0.535 NeuralNetwork 0.597 0.633 0.636 0.632 0.373 RandomForest 0.747 0.484 0.909 0.396 0.377 Xgboost 0.659 0.703 0.591 0.726 0.406 KNN 0.705 0.734 0.636 0.755 0.452 Adaboost 0.613 0.695 0.455 0.745 0.339 LightGBM 0.652 0.781 0.591 0.821 0.481 CatBoost 0.767 0.812 0.636 0.849 0.538 Abbreviations:l ogistic Regression (LR);K-Nearest Neighbors (KNN);Support Vector Machine (SVM); Gradient Boosting Machine (GBM);Extreme Gradient Boosting (XGBoost); Adaptive Boosting (AdaBoost);Light Gradient Boosting Machine (LightGBM);Categorical Boosting (CatBoost). 3.2 Variable Screening Using Boruta Algorithm for Model Inclusion For variable screening, we selected the Boruta algorithm for feature selection. The x-axis represents variable names, and the y-axis indicates the Z-scores of each variable. Boxplots display the Z-scores of variables during model computation. Green boxes denote important variables, yellow boxes represent acceptable variables, and red boxes indicate unimportant variables. Variables filtered through green and yellow criteria—including myoglobin, diabetes, cerebrovascular disease history, systolic blood pressure, aortic cross-clamp time exceeding 3 hours, intraoperative blood loss, diastolic blood pressure, cardiopulmonary bypass time, and age—were incorporated into multivariate logistic regression. Their AUC values are presented (Fig. 3 ), and Variables with P < 0.1 in multivariate logistic regression were selected to construct the model. ( TABLE2 ). 3.3. Performance comparison of ten ML algorithms Ten machine learning models were compared for their performance in predicting postoperative ischemic stroke in ATAAD patients, with external validation conducted on an independent dataset separate from the training set, showing their ROC and DCA curves results (Fig. 5 A,B). In the test cohort, the GBM model exhibited outstanding predictive performance with an AUC of 0.804, indicating high accuracy. In comparison, other models showed the following AUC values: RF 0.656, XGBoost 0.659, NNET 0.597, LR 0.765, KNN 0.7085, Adaboost 0.613, LightGBM 0.652, CatBoost 0.767, while the SVM model performed worst with an AUC of 0.505 (Fig. 5 A). DCA evaluated the clinical utility of the models, further confirming that the GBM model provided the highest net benefit across most threshold probabilities, particularly in intermediate ranges, highlighting its superiority in ischemic stroke prediction (Fig. 3 B). More detailed performance metrics for each model, including sensitivity, specificity, AUC, accuracy, and F1 score, are provided ( TABLE3 ). The GBM model demonstrated the highest performance with an accuracy of 0.747, sensitivity of 0.864, specificity of 0.717, and F1 score of 0.535, indicating strong capability in distinguishing ischemic stroke occurrence post-TAAD surgery. FIGURE6 displays the confusion matrix under the GBM model. 3.4. Interpretability analysis To better understand the relationship between the model and data, we used SHAP to provide a more intuitive interpretation of the top-performing GBM model, illustrating how these variables influence ischemic stroke occurrence in the model. Figure 7 A demonstrates the five evaluated risk factors through SHAP values. The SHAP value on the x-axis is a unified metric determining how a specific feature impacts the model's outcome. In each feature importance row, participants' attribution to outcomes is plotted as color-coded dots, with yellow and purple points representing high-risk and low-risk values, respectively. Figure 7 B displays important features in the model, where the feature ranking on the y-axis indicates their predictive importance. Results revealed strong correlations between cardiopulmonary bypass time, intraoperative blood loss, cerebrovascular disease history, systolic blood pressure, age, and predicted ischemic stroke probability. We also employed SHAP dependence plot (Fig. 7 C) to assess nonlinear feature effect. Additionally, we provided a ischemic stroke-free prediction example (Fig. 7 D) to demonstrate model interpretability. 3.5: Linear relationship between continuous variables in the model and postoperative ischemic stroke in TAAD patients We used Rcs curves to explore the linear relationship between continuous variables in the model and postoperative ischemic stroke in ATAAD patients (Fig. 8 ). The results showed that age (P = 0.007), cardiopulmonary bypass time (P < 0.001), and intraoperative blood loss (P 1 using Rcs curves, revealing increased ischemic stroke probability when CPBT exceeded 253.5min, intraoperative blood loss surpassed 2442.85ml, age was above 52 years. Discussion In this study, we compared multiple machine learning models to evaluate their performance in predicting postoperative ischemic stroke in patients undergoing Sun's procedure. Based on comprehensive analysis of multiple metrics including ROC, DCA, accuracy, sensitivity, specificity, and F1 score, the GBM model was selected due to its superior overall performance and ability to effectively balance predictive accuracy with clinical utility( TABLE3 ). These metrics highlight GBM's capability to balance sensitivity and specificity, minimizing missed sepsis cases while maintaining prediction accuracy. Through Boruta algorithm and AUC value analysis, key features including cardiopulmonary bypass time, age, intraoperative blood loss, systolic blood pressure, and cerebrovascular disease history were identified as significant predictors of postoperative ischemic stroke. RCS analysis revealed linear relationships and OR > 1 thresholds between these variables and postoperative ischemic stroke risk, providing valuable insights into how changes in these parameters influence clinical outcomes. To our knowledge, this study represents the first instance of utilizing machine learning algorithms with real-world data to predict postoperative ischemic stroke in patients undergoing Sun's procedure. However, previous studies have extensively analyzed risk factors for postoperative ischemic stroke in aortic dissection patients, many of which overlap with variables incorporated into our model. Prolonged cardiopulmonary bypass time as an independent risk factor for ischemic stroke in aortic dissection surgery has been reported in multiple studies [14] . In our findings, cardiopulmonary bypass time (OR = 1.01, 95% CI:1.00-1.02, p = 0.007) emerged as a risk factor for ischemic stroke, consistent with Mircea [14] and GERAADA [15] studies, which identified prolonged bypass time as an independent risk factor for neurological complications. Additionally, RCS curve analysis revealed that when bypass time exceeded 253.7min, ischemic stroke probability significantly increased. A study of 501 aortic arch surgery patients demonstrated that permanent neurological dysfunction correlated with prolonged operative time, while hypothermic circulatory arrest duration associated with transient neurological deficits [16] . Furthermore, the impact of cannulation strategies on postoperative ischemic stroke remains debated. Multiple studies suggest axillary artery cannulation [9, 17–19] exhibits a trend toward reduced ischemic stroke risk by enabling antegrade cerebral perfusion during arch repair. Conversely, femoral artery cannulation introduces retrograde flow through the descending aorta, potentially mobilizing arterial embolism and increasing ischemic stroke risk [19, 20] . Age is another critical risk factor for ischemic stroke prediction. Lin et al [21] reported a study of 1,445 TAAD patients undergoing TAR + FET, finding that advanced age correlates with adverse postoperative outcomes in aortic dissection, with a mean patient age of 47.11 ± 9.99 years. Our cohort exhibited a median age of 48.00 (40.00, 56.00) years, aligning with these findings. Tamura et al [22] conducted a retrospective ATAAD study demonstrating that TAR + FET is more frequently selected for individuals under 50 years. Philip et al., analyzing the IRAD database, revealed that TAAD patients aged ≥ 70 years face significantly increased in-hospital mortality risk from cerebrovascular events, though long-term mortality in survivors remains unaffected. Chinese ATAAD patients are notably younger than Western populations. The GERAADA study [15] reported a mean age of 61.3 ± 13.5 years, while the International Registry of Acute Aortic Dissection(IRAD) cohort averaged 61 ± 14.6 years [23] . Given China's higher life expectancy and emphasis on long-term outcomes, TAR + FET has become the standardized surgical approach for ATAAD involving the entire aortic arch and descending aorta, widely adopted domestically. Research on blood pressure causing ischemic stroke is still at a rather contradictory stage, and more studies are needed to explore this issue. In addition, we included some predictive indicators that have been rarely reported in the past into the model. Through multivariate logistic regression, we found that IBLV, SBP and preoperative cerebrovascular history were also risk factors for predicting postoperative ischemic stroke in patients. There have been few reports on the impact of the magnitude of preoperative systolic blood pressure on postoperative ischemic stroke in patients with dissection. Zhao et al [24] reported that hypotension upon admission (odds ratio: 9.644; P = 0.016) could also lead to new-onset postoperative ischemic stroke. Mohammed et al [25] reported that hypertension was proven to be a protective factor against long-term mortality (HR: 0.59, 95% CI [0.43–0.82], p = 0.001), and hypotension before cardiopulmonary bypass (mean arterial pressure ≤ 50 mmHg; OR: 2.17, 95% CI: 1.06–4.44, P = 0.035) was an important risk factor for patient mortality. Sabrina et al. [26] reported that uncontrolled hypertension was associated with postoperative complications and poor outcomes after endovascular aneurysm repair. Besides, our study found that excessive intraoperative blood loss (surpassing 2442.85 mL) can lead to the development of ischemic stroke. This is likely attributable to the intraoperative administration of concentrated red blood cells to maintain the patient's blood volume, along with the application of certain coagulation factors for hemostasis. Numerous studies have reported that the intraoperative transfusion of 1 to 2 units of concentrated red blood cells is associated with an increased risk of stroke [27–30] . Martin et al reported [31] that administering rFVIIa (recombinant factor VIIa) in cardiac surgery patients may lead to a significant increase in stroke [ OR = 3.69 [1.1-12.38], p = 0.03 ) ]. Similarly, Yuji Kanaoka et al reported [32] that an intraoperative blood loss ≥ 800mL was an independent risk factor for cerebral infarction ( OR = 24.31, P = 0.017) in a cohort of439 patients undergoing TEVAR for aortic aneurysm. In this study, GBM demonstrated clear superiority over traditional methods like logistic regression in predicting postoperative ischemic stroke risk in patients undergoing TAR + FET. While logistic regression is valued for its simplicity and interpretability, its reliance on linear assumptions and limited capacity to capture complex interactions among predictors constrained its performance. In contrast, GBM achieved higher accuracy, recall, and F1 scores due to its ability to model nonlinear relationships and interactions. Additionally, the use of SHAP analysis addressed GBM’s interpretability limitations, providing meaningful insights into the importance of individual predictors, thereby enhancing its clinical utility. This study has several limitations. First, retrospective studies are prone to selection bias. Second, the lack of external validation for the model and limited dataset size restrict generalizability—external validation using multicenter data from diverse geographic regions is essential to ensure robustness and broader applicability. Collaborations with other institutions are underway to address these limitations and improve clinical utility across patient populations. Finally, critical clinical data such as cardiopulmonary bypass arterial cannulation methods and temperature during deep hypothermic circulatory arrest were insufficiently documented. Future studies should aim to resolve these limitations to further refine the model. Declarations statement of authorship : This author takes responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation Funding : Beijing Natural Science Foundation Project - Joint Fund will pay for the publication of this article. Funding Number:L241032 CRediT authorship contribution statement: Siji Chen: Writing – review & editing, Writing – original draft, Project administration, Methodology, Formal analysis, Data curation, Conceptualization. WenJian Ma: Writing – original draft, Methodology, Data curation. Yang Zhao: Investigation, Data curation. Shuanglei Zhao: Investigation, Data curation. Qianxian Li, Hu Yi: Methodology, Data curation. Ming Gong: Writing – review & editing, Writing – original draft, Supervision, Methodology, Investigation, Funding acquisition, Data curation Declaration of competing interest : The authors declare that they have no known competing financial or non-financial interests that could have influenced the work reported in this paper. Human Ethics and Consent to Participate declarations : This retrospective study was approved by the Ethics Committee of Beijing Anzhen Hospital (Approval No:2025124X) with a waiver of the requirement for informed consent from the study participants. All analyses were conducted on anonymized/de-identified data, strictly protecting participant privacy Consent to Participate declaration: All investigators have consented to participate in this research, with signed documents attached as appendix Availability of Data and Materials ： All the data and materials are genuine and reliable.For the original data, please contact the author Siji Chen at [email protected] Clinical trail number : no applicable Consent for publication : Not Applicable. References Milewicz DM, Ramirez F: Therapies for Thoracic Aortic Aneurysms and Acute Aortic Dissections . Arterioscler Thromb Vasc Biol 2019, 39 (2):126-136. Ma WG, Zheng J, Dong SB, Lu W, Sun K, Qi RD, Liu YM, Zhu JM, Chang Q, Sun LZ: Sun's procedure of total arch replacement using a tetrafurcated graft with stented elephant trunk implantation: analysis of early outcome in 398 patients with acute type A aortic dissection . Ann Cardiothorac Surg 2013, 2 (5):621-628. Hagan PG, Nienaber CA, Isselbacher EM, Bruckman D, Karavite DJ, Russman PL, Evangelista A, Fattori R, Suzuki T, Oh JK et al : The International Registry of Acute Aortic Dissection (IRAD): new insights into an old disease . Jama 2000, 283 (7):897-903. Dumfarth J, Kofler M, Stastny L, Plaikner M, Krapf C, Semsroth S, Grimm M: Stroke after emergent surgery for acute type A aortic dissection: predictors, outcome and neurological recovery . Eur J Cardiothorac Surg 2018, 53 (5):1013-1020. Ghoreishi M, Sundt TM, Cameron DE, Holmes SD, Roselli EE, Pasrija C, Gammie JS, Patel HJ, Bavaria JE, Svensson LG et al : Factors associated with acute stroke after type A aortic dissection repair: An analysis of the Society of Thoracic Surgeons National Adult Cardiac Surgery Database . J Thorac Cardiovasc Surg 2020, 159 (6):2143-2154.e2143. Robu M, Marian DR, Margarint I, Radulescu B, Știru O, Iosifescu A, Voica C, Cacoveanu M, Ciomag Ianula R, Gașpar BS et al : Association between Bilateral Selective Antegrade Cerebral Perfusion and Postoperative Ischemic Stroke in Patients with Emergency Surgery for Acute Type A Aortic Dissection-Single Centre Experience . Medicina (Kaunas) 2023, 59 (8). Cofre-Martel S, Lopez Droguett E, Modarres M: Big Machinery Data Preprocessing Methodology for Data-Driven Models in Prognostics and Health Management . Sensors (Basel) 2021, 21 (20). Al'Aref SJ, Anchouche K, Singh G, Slomka PJ, Kolli KK, Kumar A, Pandey M, Maliakal G, van Rosendael AR, Beecy AN et al : Clinical applications of machine learning in cardiovascular disease and its relevance to cardiac imaging . Eur Heart J 2019, 40 (24):1975-1986. Jiang F, Jiang Y, Zhi H, Dong Y, Li H, Ma S, Wang Y, Dong Q, Shen H, Wang Y: Artificial intelligence in healthcare: past, present and future . Stroke Vasc Neurol 2017, 2 (4):230-243. Fralick M, Colak E, Mamdani M: Machine Learning in Medicine . N Engl J Med 2019, 380 (26):2588-2589. Erbel R, Aboyans V, Boileau C, Bossone E, Bartolomeo RD, Eggebrecht H, Evangelista A, Falk V, Frank H, Gaemperli O et al : 2014 ESC Guidelines on the diagnosis and treatment of aortic diseases: Document covering acute and chronic aortic diseases of the thoracic and abdominal aorta of the adult. The Task Force for the Diagnosis and Treatment of Aortic Diseases of the European Society of Cardiology (ESC) . Eur Heart J 2014, 35 (41):2873-2926. Bushnell C, Kernan WN, Sharrief AZ, Chaturvedi S, Cole JW, Cornwell WK, 3rd, Cosby-Gaither C, Doyle S, Goldstein LB, Lennon O et al : 2024 Guideline for the Primary Prevention of Stroke: A Guideline From the American Heart Association/American Stroke Association . Stroke 2024, 55 (12):e344-e424. Sun L, Qi R, Chang Q, Zhu J, Liu Y, Yu C, Zhang H, Lv B, Zheng J, Tian L et al : Surgery for marfan patients with acute type a dissection using a stented elephant trunk procedure . Ann Thorac Surg 2008, 86 (6):1821-1825. Robu M, Margarint IM, Robu C, Hanganu A, Radulescu B, Stiru O, Iosifescu A, Preda S, Cacoveanu M, Voica C et al : Factors Associated with Newly Developed Postoperative Neurological Complications in Patients with Emergency Surgery for Acute Type A Aortic Dissection . Medicina (Kaunas) 2023, 60 (1). Conzelmann LO, Hoffmann I, Blettner M, Kallenbach K, Karck M, Dapunt O, Borger MA, Weigang E: Analysis of risk factors for neurological dysfunction in patients with acute aortic dissection type A: data from the German Registry for Acute Aortic Dissection type A (GERAADA) . Eur J Cardiothorac Surg 2012, 42 (3):557-565. Khaladj N, Shrestha M, Meck S, Peterss S, Kamiya H, Kallenbach K, Winterhalter M, Hoy L, Haverich A, Hagl C: Hypothermic circulatory arrest with selective antegrade cerebral perfusion in ascending aortic and aortic arch surgery: a risk factor analysis for adverse outcome in 501 patients . J Thorac Cardiovasc Surg 2008, 135 (4):908-914. Patris V, Toufektzian L, Field M, Argiriou M: Is axillary superior to femoral artery cannulation for acute type A aortic dissection surgery? Interact Cardiovasc Thorac Surg 2015, 21 (4):515-520. Svensson LG, Blackstone EH, Rajeswaran J, Sabik JF, 3rd, Lytle BW, Gonzalez-Stawinski G, Varvitsiotis P, Banbury MK, McCarthy PM, Pettersson GB et al : Does the arterial cannulation site for circulatory arrest influence stroke risk? Ann Thorac Surg 2004, 78 (4):1274-1284; discussion 1274-1284. Gulbins H, Pritisanac A, Ennker J: Axillary versus femoral cannulation for aortic surgery: enough evidence for a general recommendation? Ann Thorac Surg 2007, 83 (3):1219-1224. Hedayati N, Sherwood JT, Schomisch SJ, Carino JL, Markowitz AH: Axillary artery cannulation for cardiopulmonary bypass reduces cerebral microemboli . J Thorac Cardiovasc Surg 2004, 128 (3):386-390. Lin H, Chang Y, Zhou H, Li J, Zhou C, Huo X: Early results of frozen elephant trunk in acute type-A dissection in 1445 patients . Int J Cardiol 2023, 389 :131213. Tamura K, Chikazawa G, Hiraoka A, Totsugawa T, Yoshitaka H: Characteristics and Surgical Results of Acute Type A Aortic Dissection in Patients Younger Than 50 Years of Age . Ann Vasc Dis 2019, 12 (4):507-513. Evangelista A, Isselbacher EM, Bossone E, Gleason TG, Eusanio MD, Sechtem U, Ehrlich MP, Trimarchi S, Braverman AC, Myrmel T et al : Insights From the International Registry of Acute Aortic Dissection: A 20-Year Experience of Collaborative Clinical Research . Circulation 2018, 137 (17):1846-1860. Zhao H, Li C, Duan W, Wei D, Xue R, Wei M, Chang Y, Shang L, Lin S, Xu J et al : Neurological prognosis in surgically treated acute aortic dissection with brain computed tomography perfusion . Eur J Cardiothorac Surg 2024, 65 (1). Al-Tawil M, Salem M, Friedrich C, Diraz S, Broll A, Rezahie N, Schoettler J, de Silva N, Puehler T, Cremer J et al : Preoperative Imaging Signs of Cerebral Malperfusion in Acute Type A Aortic Dissection: Influence on Outcomes and Prognostic Implications-A 20-Year Experience . J Clin Med 2023, 12 (20). Straus S, Farah M, Pillai K, Siracuse JJ, Alsaigh T, Malas M: Uncontrolled hypertension is associated with complications and poorer outcomes after endovascular aneurysm repair . J Vasc Surg 2025, 81 (3):606-612. Valentijn TM, Hoeks SE, Bakker EJ, van de Luijtgaarden KM, Verhagen HJ, Stolker RJ, van Lier F: The impact of perioperative red blood cell transfusions on postoperative outcomes in vascular surgery patients . Annals of vascular surgery 2015, 29 (3):511-519. Rubinstein C, Davenport DL, Dunnagan R, Saha SP, Ferraris VA, Xenos ES: Intraoperative blood transfusion of one or two units of packed red blood cells is associated with a fivefold risk of stroke in patients undergoing elective carotid endarterectomy . Journal of vascular surgery 2013, 57 (2 Suppl):53s-57s. Mariscalco G, Biancari F, Juvonen T, Zanobini M, Cottini M, Banach M, Murphy GJ, Beghi C, Angelini GD: Red blood cell transfusion is a determinant of neurological complications after cardiac surgery . Interactive cardiovascular and thoracic surgery 2015, 20 (2):166-171. Kim Y, Spolverato G, Lucas DJ, Ejaz A, Xu L, Wagner D, Frank SM, Pawlik TM: Red Cell Transfusion Triggers and Postoperative Outcomes After Major Surgery . Journal of gastrointestinal surgery : official journal of the Society for Surgery of the Alimentary Tract 2015, 19 (11):2062-2073. Ponschab M, Landoni G, Biondi-Zoccai G, Bignami E, Frati E, Nicolotti D, Monaco F, Pappalardo F, Zangrillo A: Recombinant activated factor VII increases stroke in cardiac surgery: a meta-analysis . Journal of cardiothoracic and vascular anesthesia 2011, 25 (5):804-810. Kanaoka Y, Ohki T, Maeda K, Baba T, Fujita T: Multivariate Analysis of Risk Factors of Cerebral Infarction in 439 Patients Undergoing Thoracic Endovascular Aneurysm Repair . Medicine 2016, 95 (15):e3335. Additional Declarations No competing interests reported. Supplementary Files dataall.csv Cite Share Download PDF Status: Published Journal Publication published 10 Dec, 2025 Read the published version in BMC Cardiovascular Disorders → Version 1 posted Editorial decision: Revision requested 15 Oct, 2025 Reviews received at journal 12 Oct, 2025 Reviews received at journal 02 Oct, 2025 Reviewers agreed at journal 02 Oct, 2025 Reviewers agreed at journal 26 Sep, 2025 Reviews received at journal 13 Sep, 2025 Reviews received at journal 28 Aug, 2025 Reviewers agreed at journal 19 Aug, 2025 Reviewers agreed at journal 14 Aug, 2025 Reviewers invited by journal 14 Aug, 2025 Editor assigned by journal 13 Aug, 2025 Editor invited by journal 12 Aug, 2025 Submission checks completed at journal 12 Aug, 2025 First submitted to journal 12 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-7277000","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":502683759,"identity":"12759fcc-81da-4faf-90a6-feb62db406e8","order_by":0,"name":"WenJian Ma","email":"","orcid":"","institution":"Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"WenJian","middleName":"","lastName":"Ma","suffix":""},{"id":502683760,"identity":"7c0601ae-cb00-4d46-b8b8-37d183763e34","order_by":1,"name":"Siji Chen","email":"","orcid":"","institution":"Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Siji","middleName":"","lastName":"Chen","suffix":""},{"id":502683761,"identity":"91f04b9c-9b9f-49e1-b229-59331ec397b5","order_by":2,"name":"Yang Zhao","email":"","orcid":"","institution":"Shenyang Medical College","correspondingAuthor":false,"prefix":"","firstName":"Yang","middleName":"","lastName":"Zhao","suffix":""},{"id":502683762,"identity":"4f57fc71-f7dd-43c5-9034-d96a23ad980b","order_by":3,"name":"Shuanglei Zhao","email":"","orcid":"","institution":"Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Shuanglei","middleName":"","lastName":"Zhao","suffix":""},{"id":502683763,"identity":"ed2af5ef-85c4-443f-8db3-a9723f44d22f","order_by":4,"name":"Qianxian Li","email":"","orcid":"","institution":"Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Qianxian","middleName":"","lastName":"Li","suffix":""},{"id":502683764,"identity":"34ffad4a-e6ef-4e91-9c5b-7dc900ad33fc","order_by":5,"name":"Yi Hu","email":"","orcid":"","institution":"Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yi","middleName":"","lastName":"Hu","suffix":""},{"id":502683765,"identity":"d0a5b628-4581-42ea-86da-bda35e636229","order_by":6,"name":"Ming Gong","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA+0lEQVRIiWNgGAWjYNACAwYGfmb+j4/BHGbmBuK0SLYzGBuDWcyMxGgB6TrPYCYN1sJAQIu8e+/hVzcK7tjNbGZIqy6o+BPN3w7U8qNiG04thmfOpVnnGDxL7mdmOHZ7xhmD3BmHGRsYe87cxq1lRo6ZcY7B4WTJZsa227xtBrkNQC3MQDZhLQaHmdmKQVrmE9IiL5Fj/Bioxc7gMBsbM0jLBkJaDHjOmDEDtSRINvMwS/OcMc7dCNRyEJ9f5Nt7jD/n/Dlsz89/hvEzT4Vc7rzzhw8++FGBx5YDDGwSQDqxAVn0AE71IFsaGJg/AGl7fIpGwSgYBaNghAMAXvJX+AxcDLsAAAAASUVORK5CYII=","orcid":"","institution":"Capital Medical University","correspondingAuthor":true,"prefix":"","firstName":"Ming","middleName":"","lastName":"Gong","suffix":""}],"badges":[],"createdAt":"2025-08-02 09:08:21","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7277000/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7277000/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12872-025-05375-3","type":"published","date":"2025-12-10T15:57:28+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":89574105,"identity":"c8e2554b-22b3-44bf-bb69-eec88320245c","added_by":"auto","created_at":"2025-08-21 12:50:23","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":968043,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart of model construction\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7277000/v1/c6cfea448766413d45bf5ca1.png"},{"id":89577516,"identity":"7bde97e2-25b7-4123-8a8f-55064c492c8b","added_by":"auto","created_at":"2025-08-21 13:22:23","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":728482,"visible":true,"origin":"","legend":"\u003cp\u003eThe correlation heatmap illustrated the correlation between each factor\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7277000/v1/bf637790c149ed1733d6e8a0.png"},{"id":89574111,"identity":"2865bccf-6912-45d9-a501-0fd658be2be3","added_by":"auto","created_at":"2025-08-21 12:50:23","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":304059,"visible":true,"origin":"","legend":"\u003cp\u003eThe AUC values of each variable for predicting the outcome are presented\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7277000/v1/e90b8f93c6b2a58ce63be196.png"},{"id":89575610,"identity":"bcbe7f82-1c1d-48ca-b097-f82209c938b1","added_by":"auto","created_at":"2025-08-21 12:58:23","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":3693107,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e(A)\u003c/strong\u003eReceiver Operating Characteristic (ROC) curves for the test cohorts. The curves illustrate the discriminatory ability of different predictive models, with the area under the ROC curve (AUC) values displayed for ten models.\u003cstrong\u003e(B)\u003c/strong\u003eDecision Curve Analysis (DCA) for the test cohorts. The curves show the net benefit of each model across various threshold probabilities, comparing them with the “Treat All” and “Treat None” strategies, indicating their potential clinical utility.\u003cstrong\u003e(C)\u003c/strong\u003eCalibration Curves for the test cohorts.The curves verify the degree of conformity between the apparent line and the ideal line.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-7277000/v1/4dfc10a1d3fc4e2e02e4ac51.png"},{"id":89575616,"identity":"0233ecbc-a923-4e46-87c0-10ff974c0ed3","added_by":"auto","created_at":"2025-08-21 12:58:24","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":206472,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eConfusion matrix of the GBM model for stroke prediction.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-7277000/v1/712e35b5321684167bb229c2.png"},{"id":89575608,"identity":"b824914a-63c6-4a7f-91db-948bfdddd157","added_by":"auto","created_at":"2025-08-21 12:58:23","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":1401456,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSHAP (Shapley Additive explanations) analysis of the model. (A) SHAP summary plot visualizing the distribution of SHAP values for each feature. Each dot represents an individual data point, with the x-axis showing the SHAP value (feature’s impact on the model output) and the color representing the feature value (yellow for higher values, purple for lower values).(B) Bar plot showing the mean SHAP values for the top features ranked by their contribution to stroke prediction in ATAAD patients. Features include age, history of cerebrovascular disease, cardiopulmonary bypass time(CPBT), intraoperative blood loss volume(IBLV), and preoperative systolic blood pressure(SBP) Higher mean SHAP values indicate greater importance of the feature in the model.(C).\u003c/strong\u003eSHAP dependence plot to assess nonlinear feature effect.\u003cstrong\u003e(D):\u003c/strong\u003eexample of SHAP predictions for the purpose of demonstration.\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-7277000/v1/91f0677673639ea28fbe5bb3.png"},{"id":89576085,"identity":"43121cc2-065f-41bd-baca-2270f81b435c","added_by":"auto","created_at":"2025-08-21 13:06:24","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":525034,"visible":true,"origin":"","legend":"\u003cp\u003eRestricted cubic spline (RCS) analysis of the continuous variables included in the model.\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-7277000/v1/c2ee050dfe8bd063c3f14663.png"},{"id":98243522,"identity":"40de6dac-35a8-496a-a67d-78a7f66ef0ba","added_by":"auto","created_at":"2025-12-15 16:08:23","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":10368652,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7277000/v1/526a1ca4-d11a-47b1-82df-eb98b6647393.pdf"},{"id":89577101,"identity":"a5c2f55b-9ceb-4cb6-97c7-832c61aaf21f","added_by":"auto","created_at":"2025-08-21 13:14:23","extension":"csv","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":47340,"visible":true,"origin":"","legend":"","description":"","filename":"dataall.csv","url":"https://assets-eu.researchsquare.com/files/rs-7277000/v1/99eb4a5dc89ec5ee772e058c.csv"}],"financialInterests":"No competing interests reported.","formattedTitle":"Machine learning models and restricted cubic spline were employed to analyze and predict postoperative ischemic stroke in type A aortic dissection patients","fulltext":[{"header":"Introduction","content":"\u003cp\u003eType A aortic dissection (TAAD), a life-threatening cardiovascular emergency disease characterized by end-organ mal-perfusion, exhibits marked time-dependent progression. Studies demonstrate an hourly mortality rate of 1-2% following symptom onset, with emergency surgical mortality decreasing from 5.8% (baseline) to 4.4% post-intervention within 48 hours\u003csup\u003e[1]\u003c/sup\u003e. Despite advancements in modern aortic surgery\u0026mdash;including Sun\u0026apos;s procedure\u003csup\u003e[2]\u003c/sup\u003e that integrates total arch replacement with specialized stent-graft deployment in the descending aorta, central arterial repair techniques, and neuroprotective strategies such as selective cerebral perfusion combined with moderate hypothermic circulatory arrest\u0026mdash;the incidence of postoperative neurological complication syndrome (NCs) remains persistently high at 17%-48%\u003csup\u003e[3]\u003c/sup\u003e.Postoperative NCs following emergency TAAD repair are associated with prolonged intensive care/hospital stays (ischemic stroke: 23 \u0026plusmn; 16 days vs no ischemic stroke: 17 \u0026plusmn; 18 days, P = 0.021) and morbidity\u003csup\u003e[4, 5]\u003c/sup\u003e.Ischemic stroke, the most clinically significant NCs, demonstrated a postoperative incidence of 24.8% in our prior study\u003csup\u003e[6]\u003c/sup\u003e.The development of early, precise predictive models for in-hospital neurological complications carries urgent clinical significance for optimizing decision-making and improving outcomes in this critical population.\u003c/p\u003e\n\u003cp\u003eWhile multiple risk factors associated with postoperative ischemic stroke have been identified in existing studies, there remains an urgent need for reliable, data-driven predictive models to systematically identify in-hospital ischemic stroke occurrence following surgical interventions. Furthermore, validated prediction tools specifically targeting ischemic stroke complications in patients undergoing Sun\u0026apos;s procedure (total arch replacement with frozen elephant trunk implantation) are critically lacking in both Chinese and international clinical practice. The development of a precise risk stratification model is therefore imperative to advance preoperative assessment, optimize preventive strategies, and guide therapeutic decision-making, ultimately aiming to reduce postoperative ischemic stroke incidence and improve survival rates through evidence-based interventions.\u003c/p\u003e\n\u003cp\u003eMachine learning (ML), a specialized subset of artificial intelligence (AI), enables automated extraction of clinically actionable insights for critical tasks including risk stratification, diagnostic classification, and survival prediction. ML algorithms have thus emerged as indispensable tools in biomedical research\u003csup\u003e[7]\u003c/sup\u003e, demonstrating capabilities to identify latent patterns within complex datasets and generate predictive outputs through advanced feature engineering\u003csup\u003e[8]\u003c/sup\u003e. Comparative analyses reveal ML\u0026apos;s superior performance metrics over conventional statistical methods, with successful clinical implementations\u003csup\u003e[9]\u003c/sup\u003e and real-time treatment optimization\u003csup\u003e[10]\u003c/sup\u003e.\u003c/p\u003e\n"},{"header":"Materials and Methods","content":"\u003cdiv id=\"Sec2\" class=\"Section2\"\u003e\u003ch2\u003e2.1Data Source\u003c/h2\u003e\u003cp\u003eThis retrospective study enrolled patients diagnosed with TAAD who underwent total aortic arch replacement with frozen elephant trunk implantation (Sun's procedure) at the Cardiac Surgery Center of Beijing Anzhen Hospital, Capital Medical University, between 2015 and 2020. A total of 430 consecutive cases with complete perioperative data were included. The primary endpoint was postoperative ischemic stroke occurrence. Diagnosis adhered to the 2014 ESC guidelines\u003csup\u003e[11]\u003c/sup\u003e, incorporating Stanford classification and confirmatory imaging magnetic resonance angiography (MRA) or computed tomography angiography (CTA). Exclusion criteria comprised:1.Non-surgically managed TAAD 2.Concurrent malignancies with limited life expectancy 3.Acute myocardial infarction secondary to severe myocardial mal-perfusion 4.Incomplete medical records. Postoperative ischemic stroke diagnosis followed internationally recognized ISCHEMIC STROKE guideline criteria\u003csup\u003e[12]\u003c/sup\u003e.The cohort was stratified into training (70%, n\u0026thinsp;=\u0026thinsp;301) and validation (30%, n\u0026thinsp;=\u0026thinsp;129) sets using stratified random sampling to balance baseline characteristics and mitigate overfitting risks. Ethical Compliance Approved by the Institutional Review Board of Beijing Anzhen Hospital (Approval No: 2025124X), this study strictly adhered to the ethical principles of the Declaration of Helsinki.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.2 Features Extraction\u003c/h2\u003e\u003cp\u003eThe extracted variables encompassed demographic characteristics (age, BMI), clinical profiles (hypertension, history of coronary artery disease, diabetes, smoking/alcohol use, history of cerebrovascular disease, renal insufficiency, prior cardiac surgery, NYHA functional class, acute renal dysfunction), surgical parameters (intraoperative blood loss, cardiopulmonary bypass duration, deep hypothermic circulatory arrest time, aortic cross-clamp time exceeding 3 hours), and ultrasonographic findings including dissection involvement of the left subclavian artery, innominate artery, left common carotid artery, along with true/false lumen perfusion patterns. Laboratory biomarkers comprised hepatic function indices (ALT, AST), coagulation markers (D-dimer), myocardial injury indicators (myoglobin), and renal function tests (serum creatinine). Blood samples were collected within 24 hours of admission, with the initial measurement utilized for variables requiring repeated assessments to ensure temporal consistency and minimize intervention-related confounding.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.3 Surgical Technique\u003c/h2\u003e\u003cp\u003eSun's procedure refers to total arch replacement using a four-branched vascular graft combined with specialized stent-graft implantation in the descending aorta, as technically detailed in the Annals of Thoracic Surgery\u003csup\u003e[13]\u003c/sup\u003e. Briefly, the procedure was performed under moderate hypothermia at 25\u0026deg;C with circulatory arrest. Cardiopulmonary bypass was established via right axillary artery cannulation, incorporating selective antegrade cerebral perfusion. The surgical steps included: stent-graft deployment in the descending aorta and total arch replacement with a four-branched graft. A specific sequence of vascular reconstruction was followed: proximal descending aorta anastomosis first, followed by the left carotid artery, ascending aorta, left subclavian artery, and finally the celiac artery. Early rewarming and reperfusion were initiated after completion of the distal anastomosis to minimize cerebral and coronary ischemic time.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e2.4 Model Construction and Validation\u003c/h2\u003e\u003cp\u003eFigure\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e displays the concise flowchart of predictive model construction and validation. Spearman correlation analysis was employed to investigate the interrelationships among the variables. The correlation heatmap (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e) illustrated the correlation between each factor, like the correlation coefficient between N and T is 0.36, which is less than 0.5, indicating a weak correlation, and the others are also weak correlations. Collinearity arises when two or more predictor variables exhibit strong correlation, thereby complicating the assessment of each variable\u0026rsquo;s distinct contribution to the outcome. So, we selected the most readily available variables among the collinear variables for further analysis.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003ewhile Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e displays the AUC values for all predictors. Feature selection was conducted using the Boruta algorithm to identify potential risk factors within the training dataset. Boruta\u0026rsquo;s algorithm is a method used to determine the most important features in a dataset. It identifies importance by comparing the Z value of each feature with the Z value of the corresponding \u0026ldquo;shadow feature\u0026rdquo;. In the algorithm, all real features are copied and shuffled, and then the Z value of each feature is obtained through the random forest model. Additionally, the Z values of the \u0026lsquo;shadow features\u0026rsquo; are generated by randomly shuffling the real features. Feature selection based on the Boruta algorithm. The horizontal axis is the name of each variable, and the vertical axis is the Z value of each variable. The box plot shows the Z value of each variable during model calculation. The green boxes represent important variables, and the red boxes represent unimportant variables(Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe selected significant variables were incorporated into ten distinct machine learning algorithms for model construction, including Logistic Regression (LR), K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Gradient Boosting Machine (GBM), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Artificial Neural Network (NNET), Adaptive Boosting (AdaBoost), Light Gradient Boosting Machine (LightGBM), and Categorical Boosting (CatBoost). Model performance was evaluated by selecting the algorithm achieving the highest AUC, with supplementary assessments of discriminatory power conducted through sensitivity, specificity, accuracy, and F1-score metrics. DCA curve quantified clinical utility by estimating net benefit across probability thresholds. The top-performing model underwent interpretability analysis using SHAP. Following parameter optimization on the training cohort, all model hyperparameters were fixed and externally validated on an independent validation cohort to ensure generalizability and mitigate overfitting risks.\u003c/p\u003e\u003cp\u003eThe interpretability of machine learning models remains a significant challenge in clinical applications. To elucidate how individual features contribute to predictions in our best-performing black-box model, we employed SHAP values\u0026mdash;a game theory-based approach that quantifies each feature's impact on model outputs by treating features as collaborative players. SHAP fairly attributes predictive contributions to each variable, enabling both global interpretation through mean absolute SHAP value ranking and local explanation via individual prediction decomposition. Feature importance was determined by calculating the mean absolute SHAP value across all observations. Additionally, we visualized force plots and summary plots to delineate the directional effects and magnitude ranges of key predictors, with violin plots further characterizing non-linear relationships between continuous variables and ischemic stroke risk .\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.5 Statistical Analysis\u003c/h2\u003e\u003cp\u003eThe study population exhibited complete data integrity with no missing values, obviating the need for imputation. All statistical analyses were conducted using R software version 4.4.2. Normally distributed continuous variables are presented as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation (SD), with between-group comparisons performed using Student's t-test. Non-normally distributed variables are expressed as median (interquartile range, IQR) and analyzed via the Mann-Whitney U test. Categorical variables are reported as counts (percentages), with group differences assessed using Pearson's chi-square test or Fisher's exact test as appropriate for cell frequencies.\u003c/p\u003e\u003c/div\u003e"},{"header":"Result","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003e3.1:Baseline characteristics\u003c/h2\u003e\u003cp\u003eFollowing rigorous screening, this study enrolled 430 patients, with 302 allocated to the training set and 128 to the testing set. The postoperative ischemic stroke incidence was 17% in both cohorts. The overall cohort demonstrated a median age of 48.0 years (40.0\u0026ndash;56.0), with a male predominance (81% female vs. 19% male). Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e details baseline characteristics including demographic parameters, vital signs, and laboratory indices. Comparative analysis revealed statistically significant differences (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) between ischemic stroke and non-ischemic stroke groups: Preoperative Factors: Ischemic stroke patients were older (median age 51 vs. 48 years, P\u0026thinsp;=\u0026thinsp;0.003) with higher systolic blood pressure (140 vs130 mmHg, P\u0026thinsp;=\u0026thinsp;0.004) and greater prevalence of cerebrovascular disease history (9.4% vs. 2.0%, P\u0026thinsp;=\u0026thinsp;0.017).Biochemical Profile: Elevated myoglobin levels (median 32.1vs. 43.8 ng/mL, P\u0026thinsp;=\u0026thinsp;0.012) and impaired hepatic function (AST: 22 vs. 27U/L, P\u0026thinsp;=\u0026thinsp;0.021) were observed in ischemic stroke cases. Intraoperative Metrics: Prolonged cardiopulmonary bypass time (CPBT: 201vs224, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and increased intraoperative blood loss (IBV: 1200 vs. 1500mL, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003ePatient Demographics and Baseline Characteristics in the Training Cohort\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCharacteristic\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eOverall \u003c/p\u003e\u003cp\u003eN\u0026thinsp;=\u0026thinsp;302\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNo stroke \u003c/p\u003e\u003cp\u003eN\u0026thinsp;=\u0026thinsp;249\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003estroke \u003c/p\u003e\u003cp\u003eN\u0026thinsp;=\u0026thinsp;53\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003ep-value1\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003esex, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003emale\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e56 (19%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e45 (18%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e11 (21%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003efemale\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e246 (81%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e204 (82%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e42 (79%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ehypertension, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e246 (81%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e203 (82%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e43 (81%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;0.9\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ediabete, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e19 (6.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e19 (7.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0 (0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.031\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCBDH, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e10 (3.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e5 (2.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e5 (9.4%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.017\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRenal insufficiency history, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e6 (2.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e5 (2.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1 (1.9%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;0.9\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCHD history, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e21 (7.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e15 (6.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e6 (11%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003esmoke, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e144 (48%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e122 (49%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e22 (42%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.3\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003edrink, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e76 (25%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e64 (26%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e12 (23%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCSH, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e15 (5.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e11 (4.4%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e4 (7.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.3\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNYHA class, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.3\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e178 (59%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e141 (57%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e37 (70%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eI\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0 (0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0 (0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0 (0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eII\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e64 (21%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e55 (22%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e9 (17%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIII\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e34 (11%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e31 (12%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3 (5.7%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIV\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e26 (8.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e22 (8.8%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e4 (7.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAKI, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e25 (8.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e19 (7.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e6 (11%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.4\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBrachiocephalic artery, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.4\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e143 (47%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e122 (49%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e21 (40%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTrue\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e153 (51%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e122 (49%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e31 (58%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFalse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e6 (2.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e5 (2.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1 (1.9%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLCAA, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.9\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e189 (63%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e154 (62%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e35 (66%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTrue\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e110 (36%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e92 (37%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e18 (34%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFalse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e3 (1.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3 (1.2%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0 (0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLSA, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.5\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e173 (57%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e139 (56%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e34 (64%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTrue\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e123 (41%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e105 (42%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e18 (34%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFalse\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e6 (2.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e5 (2.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1 (1.9%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLimb ischemia, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e37 (12%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e28 (11%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e9 (17%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eACCT, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.091\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;3h\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e285 (94%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e238 (96%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e47 (89%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;3h\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e17 (5.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e11 (4.4%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e6 (11%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAMI, n (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e27 (8.9%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e20 (8.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e7 (13%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.3\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge(years), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e48.00 (40.00, 56.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e48.00 (37.00, 55.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e51.00 (43.00, 62.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.003\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBMI(kg/m\u0026sup2;), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e26.12 (23.88, 28.41)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e26.00 (23.66, 28.40)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e26.12 (24.22, 29.35)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.3\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRate(bpm), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e84.00 (76.00, 93.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e84.00 (76.00, 93.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e82.00 (75.00, 96.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.7\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSBP(mmhg), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e130.50(118.00, 146.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e130.00(117.00, 144.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e140.00(127.00, 159.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.004\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDBP(mmhg), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e69.00 (58.00, 78.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e69.00 (58.00, 78.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e68.00 (54.00, 77.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLVEDD(mm), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e50.00 (45.00, 54.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e50.00 (45.00, 54.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e49.00 (44.00, 54.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.9\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLVEF(%), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e62.00 (60.00, 66.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e62.00 (60.00, 66.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e62.00 (58.00, 66.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.7\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDD(ug/mL), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2,197.00(972.00,3,373.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2,202.00(930.00,3,309.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2,138.00(1,118.00,3,409.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.4\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMB(ng/mL), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e33.35 (20.00, 66.40)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e32.10 (19.20, 60.70)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e43.80 (27.00, 100.80)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.012\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTNI(ng/mL), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.01 (0.00, 0.06)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.01 (0.00, 0.04)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.02 (0.01, 0.12)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.090\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALT(U/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e23.00 (16.00, 33.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e22.00 (16.00, 33.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e25.00 (17.00, 38.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAST(U/L), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e22.50 (18.00, 31.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e22.00 (18.00, 30.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e27.00 (20.00, 35.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.013\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCR(mg/dL), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e81.65 (69.30, 100.40)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e81.20 (69.20, 98.80)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e82.50 (72.00, 104.10)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.5\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCPBT(min), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e205.00 (180.00, 234.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e201.00 (179.00, 225.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e224.00 (198.00, 275.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSCPT(min), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e36.00 (29.00, 44.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e36.00 (29.00, 44.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e38.00 (30.00, 47.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.3\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDHCAT(min), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e22.00 (18.00, 29.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e22.00 (18.00, 28.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e24.00 (18.00, 30.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIBLV(ml), Median (Q1, Q3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1,300.00 (1,000.00, 1,600.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1,200.00 (1,000.00, 1,500.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1,500.00 (1,200.00, 2,000.00)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003csup\u003e1\u003c/sup\u003ePearson's Chi-squared test; Fisher's exact test; Wilcoxon rank sum test\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003e\u003csup\u003e2\u003c/sup\u003e\u003cb\u003eAbbreviations\u003c/b\u003e:CBDH: cerebrovascular disease history;CHD history:coronary heart disease history;CSH:cardiac surgery history;NYHA class:New York Heart Association (NYHA) Classification;SBP:systolic blood pressure;DBP:diastolic blood pressure;LVEF:Left Ventricular Ejection Fraction;DD:D-Dimer;MB:Myoglobin;TNI:Troponin I;ALT:Alanine Aminotransferase;AST:Aspartate Aminotransferase;LVEDD:Left Ventricular End-Diastolic Diameter;CR:Creatinine;CPBT:Cardiopulmonary Bypass Time;SCPT:Selective cerebral perfusion time;DHCAT:Deep hypothermic circulatory arrest time;IBLV:Intraoperative blood loss volume;AKI:Acute Kidney Injury;LCCA:\u0026nbsp;Left common carotid artery;LSA:Left subclavian artery;ACCT:Aortic cross-clamp time;BA:Brachiocephalic artery\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eMultivariate Logistic Regression Analysis for Sepsis in the Training Cohort\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCharacteristic\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003edesc\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNo (N\u0026thinsp;=\u0026thinsp;249)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eYes (N\u0026thinsp;=\u0026thinsp;53)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eOR (univariable)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eOR (multivariable)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e46.8\u0026thinsp;\u0026plusmn;\u0026thinsp;11.1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e52.1\u0026thinsp;\u0026plusmn;\u0026thinsp;11.8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.04 (1.01\u0026ndash;1.07, p\u0026thinsp;=\u0026thinsp;.003)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.04 (1.01\u0026ndash;1.07, p\u0026thinsp;=\u0026thinsp;.016)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCPBT\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e205.4\u0026thinsp;\u0026plusmn;\u0026thinsp;41.9\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e239.8\u0026thinsp;\u0026plusmn;\u0026thinsp;63.5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.01 (1.01\u0026ndash;1.02, p\u0026thinsp;\u0026lt;\u0026thinsp;.001)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.01 (1.00-1.02, p\u0026thinsp;=\u0026thinsp;.007)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eACCT\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u0026gt;\u0026thinsp;3h\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e238 (95.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e47 (88.7%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;3h\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e11 (4.4%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e6 (11.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e2.76 (0.97\u0026ndash;7.84, p\u0026thinsp;=\u0026thinsp;.056)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.18 (0.31\u0026ndash;4.54, p\u0026thinsp;=\u0026thinsp;.810)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIBLV\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1392.8\u0026thinsp;\u0026plusmn;\u0026thinsp;644.4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1824.5\u0026thinsp;\u0026plusmn;\u0026thinsp;950.3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.00 (1.00\u0026ndash;1.00, p\u0026thinsp;\u0026lt;\u0026thinsp;.001)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.00 (1.00\u0026ndash;1.00, p\u0026thinsp;=\u0026thinsp;.056)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSBP\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e131.7\u0026thinsp;\u0026plusmn;\u0026thinsp;23.4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e143.2\u0026thinsp;\u0026plusmn;\u0026thinsp;27.6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.02 (1.01\u0026ndash;1.03, p\u0026thinsp;=\u0026thinsp;.002)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.02 (1.01\u0026ndash;1.04, p\u0026thinsp;=\u0026thinsp;.007)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCHD history\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e244 (98%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e48 (90.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e5 (2%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e5 (9.4%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e5.08 (1.42\u0026ndash;18.24, p\u0026thinsp;=\u0026thinsp;.013)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e5.80 (1.09\u0026ndash;30.95, p\u0026thinsp;=\u0026thinsp;.040)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e94.8\u0026thinsp;\u0026plusmn;\u0026thinsp;265.2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e167.0\u0026thinsp;\u0026plusmn;\u0026thinsp;529.6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.00 (1.00\u0026ndash;1.00, p\u0026thinsp;=\u0026thinsp;.179)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.00 (1.00\u0026ndash;1.00, p\u0026thinsp;=\u0026thinsp;.320)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ediabete\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e230 (92.4%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e53 (100%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e19 (7.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0 (0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.00 (0.00-Inf, p\u0026thinsp;=\u0026thinsp;.986)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.00 (0.00-Inf, p\u0026thinsp;=\u0026thinsp;.984)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDBP\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e68.0\u0026thinsp;\u0026plusmn;\u0026thinsp;14.0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e68.0\u0026thinsp;\u0026plusmn;\u0026thinsp;15.6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.00 (0.98\u0026ndash;1.02, p\u0026thinsp;=\u0026thinsp;.982)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.98 (0.96\u0026ndash;1.01, p\u0026thinsp;=\u0026thinsp;.245)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"6\"\u003e\u003cb\u003eAbbreviations\u003c/b\u003e: ACCT༚Aortic cross-clamp time;CPBT:Cardiopulmonary Bypass Time;IBLV:Intraoperative blood loss volume;SBP:systolic blood pressure;CHD:history:coronary heart disease history;MB:Myoglobin;DBP:diastolic blood pressure\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003e:Comparison of Testing Cohort Results of the Machine Learning Models\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eModel\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eAUC\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eAccuracy\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSensitivity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eSpecificity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eF1\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.765\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.719\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.773\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.708\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.486\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSVM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.505\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.773\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.273\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.877\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.293\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGBM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.804\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.742\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.864\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.717\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.535\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNeuralNetwork\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.597\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.633\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.636\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.632\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.373\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRandomForest\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.747\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.484\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.909\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.396\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.377\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eXgboost\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.659\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.703\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.591\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.726\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.406\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eKNN\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.705\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.734\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.636\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.755\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.452\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAdaboost\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.613\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.695\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.455\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.745\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.339\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLightGBM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.652\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.781\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.591\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.821\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.481\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCatBoost\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.767\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.812\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.636\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.849\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.538\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"6\"\u003e\u003cb\u003eAbbreviations:l\u003c/b\u003eogistic Regression (LR);K-Nearest Neighbors (KNN);Support Vector Machine (SVM);\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd colspan=\"6\"\u003eGradient Boosting Machine (GBM);Extreme Gradient Boosting (XGBoost); Adaptive Boosting\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd colspan=\"6\"\u003e(AdaBoost);Light Gradient Boosting Machine (LightGBM);Categorical Boosting (CatBoost).\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\u003ch2\u003e3.2 Variable Screening Using Boruta Algorithm for Model Inclusion\u003c/h2\u003e\u003cp\u003eFor variable screening, we selected the Boruta algorithm for feature selection. The x-axis represents variable names, and the y-axis indicates the Z-scores of each variable. Boxplots display the Z-scores of variables during model computation. Green boxes denote important variables, yellow boxes represent acceptable variables, and red boxes indicate unimportant variables. Variables filtered through green and yellow criteria\u0026mdash;including myoglobin, diabetes, cerebrovascular disease history, systolic blood pressure, aortic cross-clamp time exceeding 3 hours, intraoperative blood loss, diastolic blood pressure, cardiopulmonary bypass time, and age\u0026mdash;were incorporated into multivariate logistic regression. Their AUC values are presented (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e), and Variables with P\u0026thinsp;\u0026lt;\u0026thinsp;0.1 in multivariate logistic regression were selected to construct the model. (\u003cb\u003eTABLE2\u003c/b\u003e).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003e3.3. Performance comparison of ten ML algorithms\u003c/h2\u003e\u003cp\u003eTen machine learning models were compared for their performance in predicting postoperative ischemic stroke in ATAAD patients, with external validation conducted on an independent dataset separate from the training set, showing their ROC and DCA curves results (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA,B). In the test cohort, the GBM model exhibited outstanding predictive performance with an AUC of 0.804, indicating high accuracy. In comparison, other models showed the following AUC values: RF 0.656, XGBoost 0.659, NNET 0.597, LR 0.765, KNN 0.7085, Adaboost 0.613, LightGBM 0.652, CatBoost 0.767, while the SVM model performed worst with an AUC of 0.505 (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA). DCA evaluated the clinical utility of the models, further confirming that the GBM model provided the highest net benefit across most threshold probabilities, particularly in intermediate ranges, highlighting its superiority in ischemic stroke prediction (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB). More detailed performance metrics for each model, including sensitivity, specificity, AUC, accuracy, and F1 score, are provided (\u003cb\u003eTABLE3\u003c/b\u003e). The GBM model demonstrated the highest performance with an accuracy of 0.747, sensitivity of 0.864, specificity of 0.717, and F1 score of 0.535, indicating strong capability in distinguishing ischemic stroke occurrence post-TAAD surgery. \u003cb\u003eFIGURE6\u003c/b\u003e displays the confusion matrix under the GBM model.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003e3.4. Interpretability analysis\u003c/h2\u003e\u003cp\u003eTo better understand the relationship between the model and data, we used SHAP to provide a more intuitive interpretation of the top-performing GBM model, illustrating how these variables influence ischemic stroke occurrence in the model. Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e7\u003c/span\u003eA demonstrates the five evaluated risk factors through SHAP values. The SHAP value on the x-axis is a unified metric determining how a specific feature impacts the model's outcome. In each feature importance row, participants' attribution to outcomes is plotted as color-coded dots, with yellow and purple points representing high-risk and low-risk values, respectively. Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e7\u003c/span\u003eB displays important features in the model, where the feature ranking on the y-axis indicates their predictive importance. Results revealed strong correlations between cardiopulmonary bypass time, intraoperative blood loss, cerebrovascular disease history, systolic blood pressure, age, and predicted ischemic stroke probability. We also employed SHAP dependence plot (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e7\u003c/span\u003eC) to assess nonlinear feature effect. Additionally, we provided a ischemic stroke-free prediction example (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e7\u003c/span\u003eD) to demonstrate model interpretability.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e3.5: \u003cb\u003eLinear relationship between continuous variables in the model and postoperative ischemic stroke in TAAD patients\u003c/b\u003e\u003c/h2\u003e\u003cp\u003eWe used Rcs curves to explore the linear relationship between continuous variables in the model and postoperative ischemic stroke in ATAAD patients (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003e). The results showed that age (P\u0026thinsp;=\u0026thinsp;0.007), cardiopulmonary bypass time (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and intraoperative blood loss (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) all exhibited highly significant statistical associations with the outcome. Furthermore, we identified threshold points where the OR\u0026thinsp;\u0026gt;\u0026thinsp;1 using Rcs curves, revealing increased ischemic stroke probability when CPBT exceeded 253.5min, intraoperative blood loss surpassed 2442.85ml, age was above 52 years.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, we compared multiple machine learning models to evaluate their performance in predicting postoperative ischemic stroke in patients undergoing Sun's procedure. Based on comprehensive analysis of multiple metrics including ROC, DCA, accuracy, sensitivity, specificity, and F1 score, the GBM model was selected due to its superior overall performance and ability to effectively balance predictive accuracy with clinical utility(\u003cb\u003eTABLE3\u003c/b\u003e). These metrics highlight GBM's capability to balance sensitivity and specificity, minimizing missed sepsis cases while maintaining prediction accuracy. Through Boruta algorithm and AUC value analysis, key features including cardiopulmonary bypass time, age, intraoperative blood loss, systolic blood pressure, and cerebrovascular disease history were identified as significant predictors of postoperative ischemic stroke. RCS analysis revealed linear relationships and OR\u0026thinsp;\u0026gt;\u0026thinsp;1 thresholds between these variables and postoperative ischemic stroke risk, providing valuable insights into how changes in these parameters influence clinical outcomes.\u003c/p\u003e\u003cp\u003eTo our knowledge, this study represents the first instance of utilizing machine learning algorithms with real-world data to predict postoperative ischemic stroke in patients undergoing Sun's procedure. However, previous studies have extensively analyzed risk factors for postoperative ischemic stroke in aortic dissection patients, many of which overlap with variables incorporated into our model. Prolonged cardiopulmonary bypass time as an independent risk factor for ischemic stroke in aortic dissection surgery has been reported in multiple studies\u003csup\u003e[14]\u003c/sup\u003e. In our findings, cardiopulmonary bypass time (OR\u0026thinsp;=\u0026thinsp;1.01, 95% CI:1.00-1.02, p\u0026thinsp;=\u0026thinsp;0.007) emerged as a risk factor for ischemic stroke, consistent with Mircea\u003csup\u003e[14]\u003c/sup\u003e and GERAADA\u003csup\u003e[15]\u003c/sup\u003e studies, which identified prolonged bypass time as an independent risk factor for neurological complications. Additionally, RCS curve analysis revealed that when bypass time exceeded 253.7min, ischemic stroke probability significantly increased. A study of 501 aortic arch surgery patients demonstrated that permanent neurological dysfunction correlated with prolonged operative time, while hypothermic circulatory arrest duration associated with transient neurological deficits\u003csup\u003e[16]\u003c/sup\u003e. Furthermore, the impact of cannulation strategies on postoperative ischemic stroke remains debated. Multiple studies suggest axillary artery cannulation\u003csup\u003e[9, 17\u0026ndash;19]\u003c/sup\u003e exhibits a trend toward reduced ischemic stroke risk by enabling antegrade cerebral perfusion during arch repair. Conversely, femoral artery cannulation introduces retrograde flow through the descending aorta, potentially mobilizing arterial embolism and increasing ischemic stroke risk\u003csup\u003e[19, 20]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eAge is another critical risk factor for ischemic stroke prediction. Lin et al\u003csup\u003e[21]\u003c/sup\u003e reported a study of 1,445 TAAD patients undergoing TAR\u0026thinsp;+\u0026thinsp;FET, finding that advanced age correlates with adverse postoperative outcomes in aortic dissection, with a mean patient age of 47.11\u0026thinsp;\u0026plusmn;\u0026thinsp;9.99 years. Our cohort exhibited a median age of 48.00 (40.00, 56.00) years, aligning with these findings. Tamura et al\u003csup\u003e[22]\u003c/sup\u003e conducted a retrospective ATAAD study demonstrating that TAR\u0026thinsp;+\u0026thinsp;FET is more frequently selected for individuals under 50 years. Philip et al., analyzing the IRAD database, revealed that TAAD patients aged\u0026thinsp;\u0026ge;\u0026thinsp;70 years face significantly increased in-hospital mortality risk from cerebrovascular events, though long-term mortality in survivors remains unaffected. Chinese ATAAD patients are notably younger than Western populations. The GERAADA study\u003csup\u003e[15]\u003c/sup\u003e reported a mean age of 61.3\u0026thinsp;\u0026plusmn;\u0026thinsp;13.5 years, while the International Registry of Acute Aortic Dissection(IRAD) cohort averaged 61\u0026thinsp;\u0026plusmn;\u0026thinsp;14.6 years\u003csup\u003e[23]\u003c/sup\u003e. Given China's higher life expectancy and emphasis on long-term outcomes, TAR\u0026thinsp;+\u0026thinsp;FET has become the standardized surgical approach for ATAAD involving the entire aortic arch and descending aorta, widely adopted domestically. Research on blood pressure causing ischemic stroke is still at a rather contradictory stage, and more studies are needed to explore this issue.\u003c/p\u003e\u003cp\u003eIn addition, we included some predictive indicators that have been rarely reported in the past into the model. Through multivariate logistic regression, we found that IBLV, SBP and preoperative cerebrovascular history were also risk factors for predicting postoperative ischemic stroke in patients. There have been few reports on the impact of the magnitude of preoperative systolic blood pressure on postoperative ischemic stroke in patients with dissection. Zhao et al\u003csup\u003e[24]\u003c/sup\u003ereported that hypotension upon admission (odds ratio: 9.644; P\u0026thinsp;=\u0026thinsp;0.016) could also lead to new-onset postoperative ischemic stroke. Mohammed et al\u003csup\u003e[25]\u003c/sup\u003ereported that hypertension was proven to be a protective factor against long-term mortality (HR: 0.59, 95% CI [0.43\u0026ndash;0.82], p\u0026thinsp;=\u0026thinsp;0.001), and hypotension before cardiopulmonary bypass (mean arterial pressure\u0026thinsp;\u0026le;\u0026thinsp;50 mmHg; OR: 2.17, 95% CI: 1.06\u0026ndash;4.44, P\u0026thinsp;=\u0026thinsp;0.035) was an important risk factor for patient mortality. Sabrina et al. \u003csup\u003e[26]\u003c/sup\u003ereported that uncontrolled hypertension was associated with postoperative complications and poor outcomes after endovascular aneurysm repair. Besides, our study found that excessive intraoperative blood loss (surpassing 2442.85 mL) can lead to the development of ischemic stroke. This is likely attributable to the intraoperative administration of concentrated red blood cells to maintain the patient's blood volume, along with the application of certain coagulation factors for hemostasis. Numerous studies have reported that the intraoperative transfusion of 1 to 2 units of concentrated red blood cells is associated with an increased risk of stroke\u003csup\u003e[27\u0026ndash;30]\u003c/sup\u003e. Martin et al reported\u003csup\u003e[31]\u003c/sup\u003e that administering rFVIIa (recombinant factor VIIa) in cardiac surgery patients may lead to a significant increase in stroke [\u003cem\u003eOR\u003c/em\u003e\u0026thinsp;=\u0026thinsp;3.69 [1.1-12.38], \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.03\u003cem\u003e)\u003c/em\u003e]. Similarly, Yuji Kanaoka et al reported\u003csup\u003e[32]\u003c/sup\u003ethat an intraoperative blood loss\u0026thinsp;\u0026ge;\u0026thinsp;800mL was an independent risk factor for cerebral infarction (\u003cem\u003eOR\u003c/em\u003e\u0026thinsp;=\u0026thinsp;24.31,\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.017) in a cohort of439 patients undergoing TEVAR for aortic aneurysm.\u003c/p\u003e\u003cp\u003eIn this study, GBM demonstrated clear superiority over traditional methods like logistic regression in predicting postoperative ischemic stroke risk in patients undergoing TAR\u0026thinsp;+\u0026thinsp;FET. While logistic regression is valued for its simplicity and interpretability, its reliance on linear assumptions and limited capacity to capture complex interactions among predictors constrained its performance. In contrast, GBM achieved higher accuracy, recall, and F1 scores due to its ability to model nonlinear relationships and interactions. Additionally, the use of SHAP analysis addressed GBM\u0026rsquo;s interpretability limitations, providing meaningful insights into the importance of individual predictors, thereby enhancing its clinical utility.\u003c/p\u003e\u003cp\u003eThis study has several limitations. First, retrospective studies are prone to selection bias. Second, the lack of external validation for the model and limited dataset size restrict generalizability\u0026mdash;external validation using multicenter data from diverse geographic regions is essential to ensure robustness and broader applicability. Collaborations with other institutions are underway to address these limitations and improve clinical utility across patient populations. Finally, critical clinical data such as cardiopulmonary bypass arterial cannulation methods and temperature during deep hypothermic circulatory arrest were insufficiently documented. Future studies should aim to resolve these limitations to further refine the model.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003estatement of authorship\u003c/strong\u003e: This author takes responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e: Beijing Natural Science Foundation Project - Joint Fund will pay for the publication of this article. Funding Number:L241032\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCRediT authorship contribution statement:\u003c/strong\u003eSiji Chen: Writing \u0026ndash; review \u0026amp; editing, Writing \u0026ndash; original draft, Project administration, Methodology, Formal analysis, Data curation, Conceptualization. WenJian Ma: Writing \u0026ndash; original draft, Methodology, Data curation. Yang Zhao: Investigation, Data curation. Shuanglei Zhao: Investigation, Data curation. Qianxian Li, Hu Yi: Methodology, Data curation. Ming Gong: Writing \u0026ndash; review \u0026amp; editing, Writing \u0026ndash; original draft, Supervision, Methodology, Investigation, Funding acquisition, Data curation\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of competing interest\u003c/strong\u003e: The authors declare that they have no known competing financial or non-financial interests that could have influenced the work reported in this paper.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eHuman Ethics and Consent to Participate declarations\u003c/strong\u003e: This retrospective study was approved by the Ethics Committee of \u003cstrong\u003eBeijing Anzhen Hospital (Approval No:2025124X)\u003c/strong\u003e with a waiver of the requirement for informed consent from the study participants. All analyses were conducted on anonymized/de-identified data, strictly protecting participant privacy\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to Participate declaration:\u003c/strong\u003e All investigators have consented to participate in this research, with signed documents attached as appendix\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of Data and Materials\u003c/strong\u003e\u003cstrong\u003e：\u003c/strong\u003eAll the data and materials are genuine and reliable.For the original data, please contact the author Siji Chen at [email protected]\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical trail number\u003c/strong\u003e: no applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e: Not Applicable.\u003c/p\u003e\n"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eMilewicz DM, Ramirez F: \u003cstrong\u003eTherapies for Thoracic Aortic Aneurysms and Acute Aortic Dissections\u003c/strong\u003e. \u003cem\u003eArterioscler Thromb Vasc Biol \u003c/em\u003e2019, \u003cstrong\u003e39\u003c/strong\u003e(2):126-136.\u003c/li\u003e\n\u003cli\u003eMa WG, Zheng J, Dong SB, Lu W, Sun K, Qi RD, Liu YM, Zhu JM, Chang Q, Sun LZ: \u003cstrong\u003eSun\u0026apos;s procedure of total arch replacement using a tetrafurcated graft with stented elephant trunk implantation: analysis of early outcome in 398 patients with acute type A aortic dissection\u003c/strong\u003e. \u003cem\u003eAnn Cardiothorac Surg \u003c/em\u003e2013, \u003cstrong\u003e2\u003c/strong\u003e(5):621-628.\u003c/li\u003e\n\u003cli\u003eHagan PG, Nienaber CA, Isselbacher EM, Bruckman D, Karavite DJ, Russman PL, Evangelista A, Fattori R, Suzuki T, Oh JK\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eThe International Registry of Acute Aortic Dissection (IRAD): new insights into an old disease\u003c/strong\u003e. \u003cem\u003eJama \u003c/em\u003e2000, \u003cstrong\u003e283\u003c/strong\u003e(7):897-903.\u003c/li\u003e\n\u003cli\u003eDumfarth J, Kofler M, Stastny L, Plaikner M, Krapf C, Semsroth S, Grimm M: \u003cstrong\u003eStroke after emergent surgery for acute type A aortic dissection: predictors, outcome and neurological recovery\u003c/strong\u003e. \u003cem\u003eEur J Cardiothorac Surg \u003c/em\u003e2018, \u003cstrong\u003e53\u003c/strong\u003e(5):1013-1020.\u003c/li\u003e\n\u003cli\u003eGhoreishi M, Sundt TM, Cameron DE, Holmes SD, Roselli EE, Pasrija C, Gammie JS, Patel HJ, Bavaria JE, Svensson LG\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eFactors associated with acute stroke after type A aortic dissection repair: An analysis of the Society of Thoracic Surgeons National Adult Cardiac Surgery Database\u003c/strong\u003e. \u003cem\u003eJ Thorac Cardiovasc Surg \u003c/em\u003e2020, \u003cstrong\u003e159\u003c/strong\u003e(6):2143-2154.e2143.\u003c/li\u003e\n\u003cli\u003eRobu M, Marian DR, Margarint I, Radulescu B, Știru O, Iosifescu A, Voica C, Cacoveanu M, Ciomag Ianula R, Gașpar BS\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eAssociation between Bilateral Selective Antegrade Cerebral Perfusion and Postoperative Ischemic Stroke in Patients with Emergency Surgery for Acute Type A Aortic Dissection-Single Centre Experience\u003c/strong\u003e. \u003cem\u003eMedicina (Kaunas) \u003c/em\u003e2023, \u003cstrong\u003e59\u003c/strong\u003e(8).\u003c/li\u003e\n\u003cli\u003eCofre-Martel S, Lopez Droguett E, Modarres M: \u003cstrong\u003eBig Machinery Data Preprocessing Methodology for Data-Driven Models in Prognostics and Health Management\u003c/strong\u003e. \u003cem\u003eSensors (Basel) \u003c/em\u003e2021, \u003cstrong\u003e21\u003c/strong\u003e(20).\u003c/li\u003e\n\u003cli\u003eAl\u0026apos;Aref SJ, Anchouche K, Singh G, Slomka PJ, Kolli KK, Kumar A, Pandey M, Maliakal G, van Rosendael AR, Beecy AN\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eClinical applications of machine learning in cardiovascular disease and its relevance to cardiac imaging\u003c/strong\u003e. \u003cem\u003eEur Heart J \u003c/em\u003e2019, \u003cstrong\u003e40\u003c/strong\u003e(24):1975-1986.\u003c/li\u003e\n\u003cli\u003eJiang F, Jiang Y, Zhi H, Dong Y, Li H, Ma S, Wang Y, Dong Q, Shen H, Wang Y: \u003cstrong\u003eArtificial intelligence in healthcare: past, present and future\u003c/strong\u003e. \u003cem\u003eStroke Vasc Neurol \u003c/em\u003e2017, \u003cstrong\u003e2\u003c/strong\u003e(4):230-243.\u003c/li\u003e\n\u003cli\u003eFralick M, Colak E, Mamdani M: \u003cstrong\u003eMachine Learning in Medicine\u003c/strong\u003e. \u003cem\u003eN Engl J Med \u003c/em\u003e2019, \u003cstrong\u003e380\u003c/strong\u003e(26):2588-2589.\u003c/li\u003e\n\u003cli\u003eErbel R, Aboyans V, Boileau C, Bossone E, Bartolomeo RD, Eggebrecht H, Evangelista A, Falk V, Frank H, Gaemperli O\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003e2014 ESC Guidelines on the diagnosis and treatment of aortic diseases: Document covering acute and chronic aortic diseases of the thoracic and abdominal aorta of the adult. The Task Force for the Diagnosis and Treatment of Aortic Diseases of the European Society of Cardiology (ESC)\u003c/strong\u003e. \u003cem\u003eEur Heart J \u003c/em\u003e2014, \u003cstrong\u003e35\u003c/strong\u003e(41):2873-2926.\u003c/li\u003e\n\u003cli\u003eBushnell C, Kernan WN, Sharrief AZ, Chaturvedi S, Cole JW, Cornwell WK, 3rd, Cosby-Gaither C, Doyle S, Goldstein LB, Lennon O\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003e2024 Guideline for the Primary Prevention of Stroke: A Guideline From the American Heart Association/American Stroke Association\u003c/strong\u003e. \u003cem\u003eStroke \u003c/em\u003e2024, \u003cstrong\u003e55\u003c/strong\u003e(12):e344-e424.\u003c/li\u003e\n\u003cli\u003eSun L, Qi R, Chang Q, Zhu J, Liu Y, Yu C, Zhang H, Lv B, Zheng J, Tian L\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eSurgery for marfan patients with acute type a dissection using a stented elephant trunk procedure\u003c/strong\u003e. \u003cem\u003eAnn Thorac Surg \u003c/em\u003e2008, \u003cstrong\u003e86\u003c/strong\u003e(6):1821-1825.\u003c/li\u003e\n\u003cli\u003eRobu M, Margarint IM, Robu C, Hanganu A, Radulescu B, Stiru O, Iosifescu A, Preda S, Cacoveanu M, Voica C\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eFactors Associated with Newly Developed Postoperative Neurological Complications in Patients with Emergency Surgery for Acute Type A Aortic Dissection\u003c/strong\u003e. \u003cem\u003eMedicina (Kaunas) \u003c/em\u003e2023, \u003cstrong\u003e60\u003c/strong\u003e(1).\u003c/li\u003e\n\u003cli\u003eConzelmann LO, Hoffmann I, Blettner M, Kallenbach K, Karck M, Dapunt O, Borger MA, Weigang E: \u003cstrong\u003eAnalysis of risk factors for neurological dysfunction in patients with acute aortic dissection type A: data from the German Registry for Acute Aortic Dissection type A (GERAADA)\u003c/strong\u003e. \u003cem\u003eEur J Cardiothorac Surg \u003c/em\u003e2012, \u003cstrong\u003e42\u003c/strong\u003e(3):557-565.\u003c/li\u003e\n\u003cli\u003eKhaladj N, Shrestha M, Meck S, Peterss S, Kamiya H, Kallenbach K, Winterhalter M, Hoy L, Haverich A, Hagl C: \u003cstrong\u003eHypothermic circulatory arrest with selective antegrade cerebral perfusion in ascending aortic and aortic arch surgery: a risk factor analysis for adverse outcome in 501 patients\u003c/strong\u003e. \u003cem\u003eJ Thorac Cardiovasc Surg \u003c/em\u003e2008, \u003cstrong\u003e135\u003c/strong\u003e(4):908-914.\u003c/li\u003e\n\u003cli\u003ePatris V, Toufektzian L, Field M, Argiriou M: \u003cstrong\u003eIs axillary superior to femoral artery cannulation for acute type A aortic dissection surgery?\u003c/strong\u003e \u003cem\u003eInteract Cardiovasc Thorac Surg \u003c/em\u003e2015, \u003cstrong\u003e21\u003c/strong\u003e(4):515-520.\u003c/li\u003e\n\u003cli\u003eSvensson LG, Blackstone EH, Rajeswaran J, Sabik JF, 3rd, Lytle BW, Gonzalez-Stawinski G, Varvitsiotis P, Banbury MK, McCarthy PM, Pettersson GB\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eDoes the arterial cannulation site for circulatory arrest influence stroke risk?\u003c/strong\u003e \u003cem\u003eAnn Thorac Surg \u003c/em\u003e2004, \u003cstrong\u003e78\u003c/strong\u003e(4):1274-1284; discussion 1274-1284.\u003c/li\u003e\n\u003cli\u003eGulbins H, Pritisanac A, Ennker J: \u003cstrong\u003eAxillary versus femoral cannulation for aortic surgery: enough evidence for a general recommendation?\u003c/strong\u003e \u003cem\u003eAnn Thorac Surg \u003c/em\u003e2007, \u003cstrong\u003e83\u003c/strong\u003e(3):1219-1224.\u003c/li\u003e\n\u003cli\u003eHedayati N, Sherwood JT, Schomisch SJ, Carino JL, Markowitz AH: \u003cstrong\u003eAxillary artery cannulation for cardiopulmonary bypass reduces cerebral microemboli\u003c/strong\u003e. \u003cem\u003eJ Thorac Cardiovasc Surg \u003c/em\u003e2004, \u003cstrong\u003e128\u003c/strong\u003e(3):386-390.\u003c/li\u003e\n\u003cli\u003eLin H, Chang Y, Zhou H, Li J, Zhou C, Huo X: \u003cstrong\u003eEarly results of frozen elephant trunk in acute type-A dissection in 1445 patients\u003c/strong\u003e. \u003cem\u003eInt J Cardiol \u003c/em\u003e2023, \u003cstrong\u003e389\u003c/strong\u003e:131213.\u003c/li\u003e\n\u003cli\u003eTamura K, Chikazawa G, Hiraoka A, Totsugawa T, Yoshitaka H: \u003cstrong\u003eCharacteristics and Surgical Results of Acute Type A Aortic Dissection in Patients Younger Than 50 Years of Age\u003c/strong\u003e. \u003cem\u003eAnn Vasc Dis \u003c/em\u003e2019, \u003cstrong\u003e12\u003c/strong\u003e(4):507-513.\u003c/li\u003e\n\u003cli\u003eEvangelista A, Isselbacher EM, Bossone E, Gleason TG, Eusanio MD, Sechtem U, Ehrlich MP, Trimarchi S, Braverman AC, Myrmel T\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eInsights From the International Registry of Acute Aortic Dissection: A 20-Year Experience of Collaborative Clinical Research\u003c/strong\u003e. \u003cem\u003eCirculation \u003c/em\u003e2018, \u003cstrong\u003e137\u003c/strong\u003e(17):1846-1860.\u003c/li\u003e\n\u003cli\u003eZhao H, Li C, Duan W, Wei D, Xue R, Wei M, Chang Y, Shang L, Lin S, Xu J\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eNeurological prognosis in surgically treated acute aortic dissection with brain computed tomography perfusion\u003c/strong\u003e. \u003cem\u003eEur J Cardiothorac Surg \u003c/em\u003e2024, \u003cstrong\u003e65\u003c/strong\u003e(1).\u003c/li\u003e\n\u003cli\u003eAl-Tawil M, Salem M, Friedrich C, Diraz S, Broll A, Rezahie N, Schoettler J, de Silva N, Puehler T, Cremer J\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003ePreoperative Imaging Signs of Cerebral Malperfusion in Acute Type A Aortic Dissection: Influence on Outcomes and Prognostic Implications-A 20-Year Experience\u003c/strong\u003e. \u003cem\u003eJ Clin Med \u003c/em\u003e2023, \u003cstrong\u003e12\u003c/strong\u003e(20).\u003c/li\u003e\n\u003cli\u003eStraus S, Farah M, Pillai K, Siracuse JJ, Alsaigh T, Malas M: \u003cstrong\u003eUncontrolled hypertension is associated with complications and poorer outcomes after endovascular aneurysm repair\u003c/strong\u003e. \u003cem\u003eJ Vasc Surg \u003c/em\u003e2025, \u003cstrong\u003e81\u003c/strong\u003e(3):606-612.\u003c/li\u003e\n\u003cli\u003eValentijn TM, Hoeks SE, Bakker EJ, van de Luijtgaarden KM, Verhagen HJ, Stolker RJ, van Lier F: \u003cstrong\u003eThe impact of perioperative red blood cell transfusions on postoperative outcomes in vascular surgery patients\u003c/strong\u003e. \u003cem\u003eAnnals of vascular surgery \u003c/em\u003e2015, \u003cstrong\u003e29\u003c/strong\u003e(3):511-519.\u003c/li\u003e\n\u003cli\u003eRubinstein C, Davenport DL, Dunnagan R, Saha SP, Ferraris VA, Xenos ES: \u003cstrong\u003eIntraoperative blood transfusion of one or two units of packed red blood cells is associated with a fivefold risk of stroke in patients undergoing elective carotid endarterectomy\u003c/strong\u003e. \u003cem\u003eJournal of vascular surgery \u003c/em\u003e2013, \u003cstrong\u003e57\u003c/strong\u003e(2 Suppl):53s-57s.\u003c/li\u003e\n\u003cli\u003eMariscalco G, Biancari F, Juvonen T, Zanobini M, Cottini M, Banach M, Murphy GJ, Beghi C, Angelini GD: \u003cstrong\u003eRed blood cell transfusion is a determinant of neurological complications after cardiac surgery\u003c/strong\u003e. \u003cem\u003eInteractive cardiovascular and thoracic surgery \u003c/em\u003e2015, \u003cstrong\u003e20\u003c/strong\u003e(2):166-171.\u003c/li\u003e\n\u003cli\u003eKim Y, Spolverato G, Lucas DJ, Ejaz A, Xu L, Wagner D, Frank SM, Pawlik TM: \u003cstrong\u003eRed Cell Transfusion Triggers and Postoperative Outcomes After Major Surgery\u003c/strong\u003e. \u003cem\u003eJournal of gastrointestinal surgery : official journal of the Society for Surgery of the Alimentary Tract \u003c/em\u003e2015, \u003cstrong\u003e19\u003c/strong\u003e(11):2062-2073.\u003c/li\u003e\n\u003cli\u003ePonschab M, Landoni G, Biondi-Zoccai G, Bignami E, Frati E, Nicolotti D, Monaco F, Pappalardo F, Zangrillo A: \u003cstrong\u003eRecombinant activated factor VII increases stroke in cardiac surgery: a meta-analysis\u003c/strong\u003e. \u003cem\u003eJournal of cardiothoracic and vascular anesthesia \u003c/em\u003e2011, \u003cstrong\u003e25\u003c/strong\u003e(5):804-810.\u003c/li\u003e\n\u003cli\u003eKanaoka Y, Ohki T, Maeda K, Baba T, Fujita T: \u003cstrong\u003eMultivariate Analysis of Risk Factors of Cerebral Infarction in 439 Patients Undergoing Thoracic Endovascular Aneurysm Repair\u003c/strong\u003e. \u003cem\u003eMedicine \u003c/em\u003e2016, \u003cstrong\u003e95\u003c/strong\u003e(15):e3335.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-cardiovascular-disorders","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcar","sideBox":"Learn more about [BMC Cardiovascular Disorders](http://bmccardiovascdisord.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcar/default.aspx","title":"BMC Cardiovascular Disorders","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Machine learning model, Risk factors, Ischemic stroke, Surgical treatment, Type A aortic dissection","lastPublishedDoi":"10.21203/rs.3.rs-7277000/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7277000/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBACKGROUD\u003c/strong\u003e: Ischemic stroke remains a devastating postoperative complication in Type A aortic dissection (TAAD) patients, contributing significantly to elevated mortality rates. Identifying reliable predictors for ischemic stroke risk is crucial for implementing timely clinical interventions. This study endeavors to develop and validate a machine learning-based predictive model for ischemic stroke risk stratification in TAAD patients undergoing surgical treatment.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e: This retrospective cohort study analyzed 430 TAAD patients who underwent total aortic arch replacement with frozen elephant trunk implantation at Beijing Anzhen Hospital (2015-2021). The cohort was randomly partitioned into training (70%, n=301) and validation (30%, n=129) sets. Feature selection was performed using Boruta algorithm, with variables demonstrating P\u0026lt;0.1 in univariate analysis subsequently incorporated into multivariate logistic regression. Ten machine learning models were evaluated through receiver operating characteristic (ROC) curves, decision curve analysis (DCA), and calibration plots. Model interpretability was enhanced via Shapley Additive Explanations (SHAP), while restricted cubic splines (RCS) elucidated potential non-linear/liner relationships between predictors and result.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults:\u003c/strong\u003e The GBM model demonstrated superior predictive performance compared to all other models, achieving an area under the curve (AUC) of 0.804 in the validation cohort. SHAP analysis identified the following key predictors of postoperative ischemic stroke: age, history of cerebrovascular disease, cardiopulmonary bypass time(CPBT), intraoperative blood loss volume(IBLV), and preoperative systolic blood pressure(SBP).Furthermore,RCS were independently constructed for each continuous variable to explore variable-outcome relationships. \u003cbr\u003e\n \u003cstrong\u003eConclusion: \u003c/strong\u003eThe Gradient Boosting Machine (GBM) model demonstrates the best predictive capacity for postoperative ischemic stroke in TAAD patients, offering clinicians a clinically actionable tool for early postoperative risk stratification and personalized therapeutic optimization.\u003c/p\u003e","manuscriptTitle":"Machine learning models and restricted cubic spline were employed to analyze and predict postoperative ischemic stroke in type A aortic dissection patients","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-21 12:50:18","doi":"10.21203/rs.3.rs-7277000/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-10-15T11:34:23+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-10-12T05:47:40+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-10-02T12:18:32+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"130522105264431863340200366510728141121","date":"2025-10-02T10:12:39+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"88068364189875030949715182185133651693","date":"2025-09-26T10:34:31+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-13T07:30:07+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-28T14:44:08+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"60302505241109729451864783574619179820","date":"2025-08-19T16:13:20+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"325929321230471204093829492973236886710","date":"2025-08-14T05:57:06+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-08-14T05:34:45+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-08-13T12:15:39+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-08-12T10:42:53+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-08-12T10:38:09+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Cardiovascular Disorders","date":"2025-08-12T10:34:26+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-cardiovascular-disorders","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcar","sideBox":"Learn more about [BMC Cardiovascular Disorders](http://bmccardiovascdisord.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcar/default.aspx","title":"BMC Cardiovascular Disorders","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"0fe4faa2-2083-4a2b-80f1-0c28d99a42d8","owner":[],"postedDate":"August 21st, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-12-15T16:01:14+00:00","versionOfRecord":{"articleIdentity":"rs-7277000","link":"https://doi.org/10.1186/s12872-025-05375-3","journal":{"identity":"bmc-cardiovascular-disorders","isVorOnly":false,"title":"BMC Cardiovascular Disorders"},"publishedOn":"2025-12-10 15:57:28","publishedOnDateReadable":"December 10th, 2025"},"versionCreatedAt":"2025-08-21 12:50:18","video":"","vorDoi":"10.1186/s12872-025-05375-3","vorDoiUrl":"https://doi.org/10.1186/s12872-025-05375-3","workflowStages":[]},"version":"v1","identity":"rs-7277000","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7277000","identity":"rs-7277000","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}