Optimizing Initial Vancomycin Dosing in Hospitalized Patients Using Machine Learning Approach for Enhanced Therapeutic Outcomes: Algorithm Development and Validation Study.

Autism spectrum disorder (ASD) is a multifaceted neurodevelopmental condition that affects millions of individuals globally. This study leverages the publicly available UCI Machine Learning ASD Screening Dataset, which contains behavioral questionnaire responses from children aged 4-11 years, to develop and compare multiple machine learning (ML) models for early ASD diagnosis. This study applies SHAP (SHapley Additive exPlanations) to interpret and analyze the feature importance of various attributes in the dataset. Advanced ML models like Extreme Gradient Boosting (XGBoost), Random Forest (RF), Extra Trees Classifier (ETC), Support Vector Machines (SVM), and Multilayer Perceptrons (MLPs) were implemented to evaluate and compare their performance in predicting early ASD. The developed ASD model underwent a rigorous evaluation through comprehensive performance metrics, including accuracy, precision, recall, F1-score, specificity, and Matthews correlation coefficient. Ranking-based metrics, such as area under the receiver operating characteristic curve (AUROC), provided additional insights into classification robustness, while 10-fold cross-validation ensured reliability and generalizability. Results indicate that XGBoost achieved the highest diagnostic performance (accuracy = 97.8%, AUROC = 0.99), followed closely by MLP, with RF and ETC offering balanced performance and interpretability. SHAP analysis identified key behavioral features influencing model predictions, enabling explainable AI-driven insights. These findings suggest that ML-based tools can support clinicians in identifying ASD earlier, allowing for timelier interventions in disability services, particularly in underserved populations. While results are promising, the use of secondary data and the lack of clinical validation remain limitations, highlighting the need for real-world pilot testing. These findings underscore the transformative potential of ML techniques in enhancing ASD diagnostics. Despite their promise, the results are constrained by reliance on secondary data and the absence of clinical validation, underscoring the need for real-world pilot testing. Nonetheless, incorporating these models into clinical practice could accelerate early detection, facilitate personalized interventions, and optimize resource allocation for improved ASD management outcomes.

Hypoxemia is a common complication associated with anesthesia in painless gastroscopy. With the aging of the social population, the number of cases of hypoxemia among middle-aged and elderly patients is increasing. However, tools for predicting hypoxemia in middle-aged and elderly patients are lacking. In this study, we investigated the risk factors for hypoxemia in middle-aged and elderly outpatients undergoing painless gastroscopy based on machine learning and constructed a risk prediction model. In this retrospective study, we included the data on 1,348 outpatients undergoing painless gastroscopy. In total, 26 characteristic variables, including demographic information, past medical history, and clinical data of the patients were included, and BorutaShap was used for feature selection. Five machine learning algorithm models, including logistic regression (LR), support vector machine (SVM), random forest (RF), extreme gradient boosting (XGB), and light gradient boosting machine (LightGBM), were selected. The best models were selected based on the area under the receiver operating characteristic curve (AUROC). Model feature importance was explained and analyzed using Shapley Additive Explanations (SHAP). The endpoint event of this study was considered to be hypoxemia during the procedure, defined as at least one occurrence of pulse oxygen saturation below 90% without probe misalignment or interference from the beginning of anesthesia induction to the end of painless gastroscopy. In the final cohort of 984 patients, 11% of patients (108/984) experienced hypoxemia during the painless gastroscopy procedure. The AUROCs of the five models were as follows: Logistic Regression (AUROC = 0.893, 95CI: 0.881–0.899), SVM (AUROC = 0.855, 95CI: 0.812–0.884), Random Forest (AUROC = 0.914, 95CI: 0.889–0.924), XGB (AUROC = 0.902, 95CI: 0.865–0.919), and LightGBM (AUROC = 0.891, 95CI: 0.847–0.917). Regarding the explanation of the importance of SHAP features, preoperative variables (baseline SpO2, body mass index, and micrognathia) and intraoperative variables (operating time of gastroscopy, induction dose of etomidate and propofol mixture, append anesthetic, cough, and repeated pharyngeal irritation) significantly contributed to the model. We identified eight potential risk factors related to the occurrence of hypoxemia in middle-aged and elderly patients undergoing painless gastroscopy, based on machine learning feature engineering. Among the five machine learning algorithms, RF exhibited the best predictive performance in the internal test set and had a certain degree of generalization ability in the external validation set, which indicated that the RF model was more suitable for the data framework of this study. This model was more likely to enhance the accuracy of hypoxemia prediction in middle-aged and elderly patients undergoing painless gastroscopy, and thus, it is suitable for assisting anesthesiologists in clinical decision-making.

Breast cancer continues to be a leading cause of death among women in the world. The prediction of survival outcomes based on treatment modalities, i.e., chemotherapy, hormone therapy, surgery, and radiation therapy is an essential step towards personalization in treatment planning. However, Machine Learning (ML) models may improve these predictions by investigating intricate relationships between clinical variables and survival. This study investigates the performance of several ML models to predict survival rate in patients undergoing diverse breast cancer treatments i.e., chemotherapy, hormone therapy, surgery and radiation using multiple clinical parameters. The dataset consisted of 5000 samples and turned into downloaded from Kaggle. The models assessed blanketed Support Vector Machines (SVM), K-Nearest Neighbor (KNN), AdaBoost, Gradient Boosting, Random Forest, Gaussian Naive Bayes, Logistic Regression, Extreme Gradient Boosting (XG boost), and Decision tree. Performance of the models was assessed using parameters such as Accuracy, Precision, Recall, F1-Score and Area under the Receiver Operating Characteristic Curve (AUC-ROC). SHAP (SHapley Additive exPlanations) values analysis was done to provide an explanation for the impact of a feature on model predictions using Waterfall and Beeswarm plots. Anticipated baseline (E(f(x))) were in comparison to the predictions (f(x)) for each therapy group. Performance of Gradient Boosting was shown to be the best with an Accuracy: 0.972, Precision: 0.973, Recall: 0.972, F1-Score: 0.973, and AUC-ROC Score: 0.997. Chemotherapy had a notably bad impact on survival, with an f(x) of -0.274 and an E(f(x)) of -0.025. Hormone therapy showed the maximum detrimental effect on survival, with an f(x) of -0.408. Surgical operation had an extraordinarily impartial impact (f(x) = -0.041), even as radiation therapy positively impacted survival results with an f(x) of 0.22. Gradient Boosting was the most predictive algorithm for breast cancer survival outcomes. This SHAP-primarily based analysis provides a complete knowledge of ways one-of-a-kind treatments have an effect on survival predictions in breast cancer patients. Radiation therapy indicates the most tremendous effect on survival, whilst hormone therapy reveals the maximum poor effect. Future studies need to explore personalized treatment strategies that comprise these insights to enhance patient effects.

PurposeThe aims of the study were to develop and evaluate a machine learning model with which to predict postnatal growth failure (PGF) among very low birth weight (VLBW) infants.Materials and MethodsOf 10425 VLBW infants registered in the Korean Neonatal Network between 2013 and 2017, 7954 infants were included. PGF was defined as a decrease in Z score >1.28 at discharge, compared to that at birth. Six metrics [area under the receiver operating characteristic curve (AUROC), accuracy, precision, sensitivity, specificity, and F1 score] were obtained at five time points (at birth, 7 days, 14 days, 28 days after birth, and at discharge). Machine learning models were built using four different techniques [extreme gradient boosting (XGB), random forest, support vector machine, and convolutional neural network] to compare against the conventional multiple logistic regression (MLR) model.ResultsThe XGB algorithm showed the best performance with all six metrics across the board. When compared with MLR, XGB showed a significantly higher AUROC (p=0.03) for Day 7, which was the primary performance metric. Using optimal cut-off points, for Day 7, XGB still showed better performances in terms of AUROC (0.74), accuracy (0.68), and F1 score (0.67). AUROC values seemed to increase slightly from birth to 7 days after birth with significance, almost reaching a plateau after 7 days after birth.ConclusionWe have shown the possibility of predicting PGF through machine learning algorithms, especially XGB. Such models may help neonatologists in the early diagnosis of high-risk infants for PGF for early intervention.

Using machine learning to predict outcomes following carotid endarterectomy

Machine-learning-derived online prediction models of outcomes for patients with cholelithiasis-induced acute cholangitis: development and validation in two retrospective cohorts

PurposeTo analyze the risk factors for preterm birth in patients with early-onset preeclampsia (EOPE) based on multi-algorithm machine learning and to construct a predictive model to explore the predictive value of the model.MethodsA retrospective analysis was conducted on 442 EOPE patients from a single tertiary center, divided into preterm birth (<37 weeks, n=358) and term-born (≥37 weeks, n=84) groups. Univariate analysis, random forest importance assessment, lasso regression combined with multivariate regression analysis were used for feature evaluation. Eight machine learning models were trained (70% data) and validated (30% data). A Stacking ensemble model was constructed, and SHapley Additive exPlanations (SHAP) was used for feature interpretation.ResultsThe area under the receiver operating characteristic curve (AUROC) for predicting preterm birth in EOPE patients using Logistic Regression, Gaussian Naive Bayes, Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine, Support Vector Machine (SVM), Gradient Boosting Decision Tree (GBDT), Multi-Layer Perceptron, and Elastic Net were 0.763, 0.712, 0.821, 0.832, 0.821, 0.842, 0.784, and 0.763, respectively. The Stacking model (XGBoost+GBDT+SVM) achieved superior performance (AUROC=0.865). Three independent risk factors were identified: fetal growth restriction (aOR=3.50, p = 0.047), serum cystatin C (aOR=11.27, p = 0.018), and C-reactive protein (aOR=1.37, p < 0.001). SHAP analysis revealed GBDT as the top contributor to Stacking predictions, with microalbunminuria (GBDT, XGBoost) and age (SVM) being the most influential features.ConclusionMachine learning models can serve as reliable assessment tools for predicting the risk of preterm birth in patients with EOPE. The ensemble prediction model demonstrates the best predictive performance, helping obstetricians identify high-risk patients and perform early intervention to improve perinatal outcomes.

ObjectiveInferior vena cava (IVC) filter placement is associated with important long-term complications. Predictive models for filter-related complications may help guide clinical decision-making but remain limited. We developed machine learning (ML) algorithms that predict 1-year IVC filter complications using preoperative data. MethodsThe Vascular Quality Initiative database was used to identify patients who underwent IVC filter placement between 2013 and 2024. We identified 77 preoperative demographic and clinical features from the index hospitalization when the filter was placed. The primary outcome was 1-year filter-related complications (composite of filter thrombosis, migration, angulation, fracture, and embolization or fragmentation, vein perforation, new caval or iliac vein thrombosis, new pulmonary embolism, access site thrombosis, or failed retrieval). The data were divided into training (70%) and test (30%) sets. Six ML models were trained using preoperative features with 10-fold cross-validation (Extreme Gradient Boosting, random forest, Naïve Bayes classifier, support vector machine, artificial neural network, and logistic regression). The primary model evaluation metric was area under the receiver operating characteristic curve (AUROC). Model robustness was assessed using calibration plot and Brier score. Performance was evaluated across subgroups based on age, sex, race, ethnicity, rurality, median Area Deprivation Index, planned duration of filter, landing site of filter, and presence of prior IVC filter placement. ResultsOverall, 14,476 patients underwent IVC filter placement and 584 (4.0%) experienced 1-year filter-related complications. Patients with a primary outcome were younger (59.3 ± 16.7 years vs 63.8 ± 16.0 years; P < .001) and more likely to have thrombotic risk factors including thrombophilia, prior venous thromboembolism (VTE), and family history of VTE. The best prediction model was Extreme Gradient Boosting, achieving an AUROC of 0.93 (95% confidence interval, 0.92-0.94). In comparison, logistic regression had an AUROC of 0.63 (95% confidence interval, 0.61-0.65). Calibration plot showed good agreement between predicted/observed event probabilities with a Brier score of 0.07. The top 10 predictors of 1-year filter-related complications were (1) thrombophilia, (2) prior VTE, (3) antiphospholipid antibodies, (4) factor V Leiden mutation, (5) family history of VTE, (6) planned duration of IVC filter (temporary), (7) unable to maintain therapeutic anticoagulation, (8) malignancy, (9) recent or active bleeding, and (10) age. Model performance remained robust across all subgroups. ConclusionsWe developed ML models that can accurately predict 1-year IVC filter complications, performing better than logistic regression. These algorithms have potential to guide patient selection for filter placement, counselling, perioperative management, and follow-up to mitigate filter-related complications and improve outcomes.

Development and validation of an interpretable machine learning model for cerebral small vessel disease risk assessment.

Social determinants of health (SDOH) play a critical role in the onset and progression of chronic kidney disease (CKD). Despite the well-established role of SDOH, previous studies have not fully incorporated these factors in predicting CKD in Type 2 diabetes patients. To bridge this gap, this study aimed to develop and evaluate the machine learning (ML) models that incorporate SDOH to enhance CKD risk prediction in Type 2 diabetes patients. Data were obtained from the 2023 Behavioral Risk Factor Surveillance System (BRFSS), a national survey that collects comprehensive health-related data from adults across the United States. Missing data were addressed using the K-nearest neighbor imputation method, and the Synthetic Minority Oversampling Technique (SMOTE) was applied to balance class distributions. Potential predictive features were selected using correlation coefficient analysis. The dataset was partitioned into training (80%) and testing (20%) subsets, with a 3-fold cross-validation strategy applied to the training data. Seven ML models were developed for CKD risk prediction, including logistic regression (LR), decision tree (DT), K-nearest neighbor (KNN), random forest (RF), eXtreme Gradient Boosting (XGBoost), and an artificial neural network (ANN). Model performance was evaluated using multiple metrics, including the area under the receiver operating characteristic curve (AUROC), precision, recall, F1 score, accuracy, and false positive rate. The study included 19,912 Type 2 diabetes patients (weighted sample size: 818,878), among whom 2,924 (weighted 13.92%) had CKD, and 16,988 (weighted 86.08%) did not. Over half of the CKD group (50.4%) were aged 65 or older. The proportion of female patients was higher in both groups, comprising 53.8% of the CKD group and 50.5% of the non-CKD group. Among the ML models evaluated, the RF model demonstrated the highest predictive performance for CKD, with an AUROC of 0.89 (95% CI: 0.88 - 0.90), followed by the DT model (0.84, 95% CI: 0.83 - 0.85) and XGBoost (0.83, 95% CI: 0.82 - 0.84). The RF model achieved an accuracy of 0.81 (95%CI: 0.81 - 0.81), a precision of 0.79 (95%CI: 0.79 - 0.79), a recall of 0.85 (95%CI: 0.85 - 0.85), and an F1 score of 0.82 (95%CI: 0.82 - 0.82). Additionally, the RF model exhibited strong calibration, reinforcing its reliability as a predictive tool for CKD risk in individuals with Type 2 diabetes. The study findings underscore the potential of ML models, particularly the RF model, in accurately predicting CKD among individuals with Type 2 diabetes. This approach not only enhances the precision of CKD prediction but also highlights the importance of addressing social and environmental disparities in disease prevention and management. Leveraging ML models with SDOH can lead to earlier interventions, more personalized treatment plans, and improved health outcomes for vulnerable populations. The online version contains supplementary material available at 10.1007/s40200-025-01621-9.

BackgroundAcute kidney disease (AKD) affects more than half of critically ill elderly patients with acute kidney injury (AKI), which leads to worse short-term outcomes.ObjectiveWe aimed to establish 2 machine learning models to predict the risk and prognosis of AKD in the elderly and to deploy the models as online apps.MethodsData on elderly patients with AKI (n=3542) and AKD (n=2661) from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database were used to develop 2 models for predicting the AKD risk and in-hospital mortality, respectively. Data collected from Xiangya Hospital of Central South University were for external validation. A bootstrap method was used for internal validation to obtain relatively stable results. We extracted the indicators within 24 hours of the first diagnosis of AKI and the fluctuation range of some indicators, namely delta (day 3 after AKI minus day 1), as features. Six machine learning algorithms were used for modeling; the area under the receiver operating characteristic curve (AUROC), decision curve analysis, and calibration curve for evaluating; Shapley additive explanation (SHAP) analysis for visually interpreting; and the Heroku platform for deploying the best-performing models as web-based apps.ResultsFor the model of predicting the risk of AKD in elderly patients with AKI during hospitalization, the Light Gradient Boosting Machine (LightGBM) showed the best overall performance in the training (AUROC=0.844, 95% CI 0.831-0.857), internal validation (AUROC=0.853, 95% CI 0.841-0.865), and external (AUROC=0.755, 95% CI 0.699–0.811) cohorts. In addition, LightGBM performed well for the AKD prognostic prediction in the training (AUROC=0.861, 95% CI 0.843-0.878), internal validation (AUROC=0.868, 95% CI 0.851-0.885), and external (AUROC=0.746, 95% CI 0.673-0.820) cohorts. The models deployed as online prediction apps allowed users to predict and provide feedback to submit new data for model iteration. In the importance ranking and correlation visualization of the model’s top 10 influencing factors conducted based on the SHAP value, partial dependence plots revealed the optimal cutoff of some interventionable indicators. The top 5 factors predicting the risk of AKD were creatinine on day 3, sepsis, delta blood urea nitrogen (BUN), diastolic blood pressure (DBP), and heart rate, while the top 5 factors determining in-hospital mortality were age, BUN on day 1, vasopressor use, BUN on day 3, and partial pressure of carbon dioxide (PaCO2).ConclusionsWe developed and validated 2 online apps for predicting the risk of AKD and its prognostic mortality in elderly patients, respectively. The top 10 factors that influenced the AKD risk and mortality during hospitalization were identified and explained visually, which might provide useful applications for intelligent management and suggestions for future prospective research.

<sec> <title>BACKGROUND</title> Type 2 diabetes mellitus (T2DM) is a major global health concern, affecting over 537 million adults worldwide in 2021, with projections reaching 783 million by 2045. Cancer represents a significant comorbidity in T2DM patients, accounting for an increasing proportion of hospital admissions and mortality. The complex interplay between T2DM and cancer involves shared risk factors including obesity, physical inactivity, chronic inflammation, and metabolic dysregulation. Early identification of cancer risk in T2DM patients could enable targeted preventive interventions and improve clinical outcomes. </sec> <sec> <title>OBJECTIVE</title> This study aimed to develop and validate machine learning models for predicting breast cancer and lung cancer risk in patients with type 2 diabetes using routinely collected clinical and biochemical parameters from electronic health records. </sec> <sec> <title>METHODS</title> This retrospective cohort study analyzed electronic health record data from the National Taiwan University Hospital Integrated Medical Database (NTUH-iMD), which stores patient-level information continuously from NTUH and its affiliated branches serving citizens throughout Taiwan. The study included 68,000 adult T2DM patients (≥20 years) from March 2007 to December 2019. Patients were identified using ICD-10 diagnostic codes for T2DM (E11) and cancer (C33-C34 for lung cancer, C50 for breast cancer). We collected demographic data and biochemical parameters including lipid profiles, inflammatory markers, renal function tests, liver enzymes, and glycemic control indicators. Three machine learning algorithms were implemented: eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Random Forest. Data preprocessing included regression-based imputation for missing values, Min-Max normalization, and k-fold cross-validation. Model performance was evaluated using accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC-ROC). Binary logistic regression analysis identified significant risk factors with odds ratios and 95% confidence intervals. This study was approved by the Research Ethics Committee of National Taiwan University Hospital (IRB No. 202110098RINA). </sec> <sec> <title>RESULTS</title> Among 68,000 T2DM patients, 1,714 developed breast cancer and 1,970 developed lung cancer during the study period. Student's t-tests revealed significant differences (p&lt;0.05) in multiple biochemical markers between cancer and non-cancer groups. For breast cancer prediction, body mass index (BMI), globulin (Glo), triglycerides (TG), high-sensitivity C-reactive protein (hsCRP), low-density lipoprotein (LDL), and alkaline phosphatase (ALP) emerged as significant risk factors. Specifically, BMI ≥35 kg/m² showed the highest risk (OR=1.651), followed by Glo &gt;3.5 g/dL (OR=5.131), and TG &gt;150 mg/dL (OR=1.352). For lung cancer prediction, significant factors included BMI, albumin (Alb), globulin (Glo), blood urea nitrogen (BUN), total cholesterol (T-CHO), hsCRP, platelet count (PLT), hemoglobin A1c (HbA1c), and fasting glucose (GLUAC). Notably, hsCRP &gt;1 mg/dL demonstrated the strongest association (OR=2.044), while high platelet count &gt;378×10³/µL showed an OR of 1.323. The machine learning models demonstrated robust predictive performance across both cancer types, with the best-performing model achieving an AUC-ROC of 0.786 in internal validation and 0.764 in external validation for lung cancer prediction. </sec> <sec> <title>CONCLUSIONS</title> This study successfully developed machine learning-based prediction models for breast cancer and lung cancer risk in T2DM patients using routinely available clinical data from electronic health records. The identification of specific biochemical markers as significant predictors provides clinically actionable insights for early cancer detection and prevention strategies in this high-risk population. The models demonstrated good discriminative ability and could be integrated into clinical decision support systems to facilitate timely interventions. Future research should focus on external validation in diverse populations, prospective validation studies, and integration with additional data modalities including genetic and lifestyle factors to further enhance prediction accuracy and clinical utility. </sec> <sec> <title>CLINICALTRIAL</title> no </sec>

P125. Development of a novel ensemble machine learning algorithm for prediction of complications and readmission after anterior cervical spinal fusion

Genetic testing can determine family screening strategies and has prognostic and diagnostic value in hypertrophic cardiomyopathy (HCM). However, it can also pose a significant psychosocial burden. Conventional scoring systems offer modest ability to predict genotype positivity. The aim of our study was to develop a novel prediction model for genotype positivity in patients with HCM by applying machine learning (ML) algorithms. We constructed 3 ML models using readily available clinical and cardiac imaging data of 102 patients from Columbia University with HCM who had undergone genetic testing (the training set). We validated model performance on 76 patients with HCM from Massachusetts General Hospital (the test set). Within the test set, we compared the area under the receiver operating characteristic curves (AUROCs) for the ML models against the AUROCs generated by the Toronto HCM Genotype Score (the Toronto score) and Mayo HCM Genotype Predictor (the Mayo score) using the Delong test and net reclassification improvement. Overall, 63 of the 178 patients (35%) were genotype positive. The random forest ML model developed in the training set demonstrated an AUROC of 0.92 (95% CI, 0.85-0.99) in predicting genotype positivity in the test set, significantly outperforming the Toronto score (AUROC, 0.77 [95% CI, 0.65-0.90], P=0.004, net reclassification improvement: P<0.001) and the Mayo score (AUROC, 0.79 [95% CI, 0.67-0.92], P=0.01, net reclassification improvement: P=0.001). The gradient boosted decision tree ML model also achieved significant net reclassification improvement over the Toronto score (P<0.001) and the Mayo score (P=0.03), with an AUROC of 0.87 (95% CI, 0.75-0.99). Compared with the Toronto and Mayo scores, all 3 ML models had higher sensitivity, positive predictive value, and negative predictive value. Our ML models demonstrated a superior ability to predict genotype positivity in patients with HCM compared with conventional scoring systems in an external validation test set.

Category:Ankle Arthritis; AnkleIntroduction/Purpose:Ankle arthrodesis and total ankle replacement are the most commonly performed procedures for surgical management of ankle arthritis. Arthrodesis provides effective pain relief but the rate of complications after arthrodesis is higher as it is more commonly performed in patients with comorbidities that preclude ankle replacement. Accurately risk- stratifying patients who undergo ankle arthrodesis would be of great utility, given the significant cost and morbidity associated with developing major perioperative complications. There is a paucity of accurate prediction models that can be used to pre- operatively risk-stratify patients for ankle arthrodesis. We aim to develop a machine learning (ML) algorithm for prediction of major perioperative complication after ankle arthrodesis as well as compare its performance against traditional predictive models based on logistic regression.Methods:This is a retrospective cohort study of adult patients who underwent ankle arthrodesis at any non-federal California hospital between 2015 and 2017. The primary outcome was readmission within 30 days or major perioperative complication - venous thromboembolism within 30 days, myocardial infarction within 7 days, pneumonia within 7 days, systemic infection within 7 days, surgical site bleeding within 90 days, and wound complications within 90 days. We build ML and logistic regression models that span different classes of modeling approaches: XGBoost, AdaBoost, Gradient Boosting, and Random Forest. Discrimination and calibration were assessed using area under the receiver operating characteristic curve (AUROC) and Brier score, respectively. We utilize a partial dependence function to measure the importance of an individual feature by assessing the average effect in predicted risks when its value is altered. We rank the contribution of the included variables to the prediction of adverse outcomes.Results:A total of 1,084 patients met inclusion criteria for this study. There were 131 major complications or readmission (12.1%). The optimized XGBoost algorithm demonstrates higher discrimination (AUROC: 0.707 + 0.052) compared to LR (0.691 + 0.055). The receiver operating characteristic curves for the XGBoost and logistic regression models are visualized in Figure 1. XGBoost also outperforms the three other ML models. This model was well calibrated (Brier score: 0.103 + 0.001). The variables most important for the XGBoost model include diabetes, chronic kidney disease, implant complication, and major fracture. Five of the ten most important features for XGBoost were markedly less important for the traditional logistic regression model: male sex, prior hip fracture, cardiorespiratory failure, acute renal failure, and dialysis status.Conclusion:We report a ML algorithm for prediction of major perioperative complications after ankle arthrodesis. The optimized XGBoost model is well-calibrated and demonstrates superior risk prediction to logistic regression. This tool may identify and address potentially modifiable risk factors, helping to accurately risk-stratify patients and decrease likelihood of major complications. Notably, the predictors most important for XGBoost are different from those for logistic regression. This suggests that the superior discriminative capability of ML methods stems from their ability to capture complex non-linear relationships between variables that logistic regression is unable to detect.