Predicting Preoperative Deep Vein Thrombosis in Elderly Hip Fracture Patients Using an Interpretable Machine Learning Model.

Objective: Deep vein thrombosis (DVT) frequently occurs in the lower extremities of elderly hip - fracture patients. This study aims to develop an interpretable machine - learning model for predicting preoperative DVT risk in these patients and use the SHapley Additive exPlanations (SHAP) method to explain the model and identify significant factors. Methods: A total of 976 patients (38 variables) were included. The dataset was randomly split into a training set (N = 683) and a validation set (N = 293). The Synthetic Minority Over - sampling Technique (SMOTE) was used to balance the training set. Logistic Regression (LR), Random Forest (RF), and Adaptive Boosting (AdaBoost) were applied to select influential factors, and Venn analysis was used to identify key variables. Five machine - learning techniques, including Extreme Gradient Boosting (XGBoost), were used to develop a predictive model. The performance of various models was evaluated to find the optimal algorithm, and the SHAP method was used for interpretation. Results: A total of eight variables were selected as inputs for the predictive model. The XGBoost model achieved the highest performance on the training set data, with an Area Under the Curve (AUC), accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and F1 score of 0.975, 0.923, 0.936, 0.910, 0.909, 0.939, and 0.922, respectively. Furthermore, the calibration curve demonstrated a high level of agreement between the predicted probabilities and the observed risks, while the decision curve revealed that the XGBoost model had a higher net benefit compared to other machine learning models. Additionally, the use of the SHAP tool facilitated the interpretation of both the features and individual predictions. Conclusion: Interpretable predictive models can help implement timely interventions and assist physicians in accurately predicting preoperative DVT risk in elderly hip - fracture patients.