Constructing a risk prediction model for post-TKA DVT formation based on machine learning methods.

Area of Science:

• Orthopedic Surgery
• Cardiovascular Medicine
• Data Science in Healthcare

Background:

• Deep vein thrombosis (DVT) is a significant complication following total knee arthroplasty (TKA).
• Accurate risk prediction models are crucial for effective DVT prevention strategies post-TKA.

Purpose of the Study:

• To construct and compare multiple machine learning models for predicting DVT risk after TKA.
• To identify key risk factors associated with post-TKA DVT.
• To determine the optimal predictive model for clinical application.

Main Methods:

• Retrospective analysis of 1238 TKA patient records.
• Utilized Lasso and Boruta for feature selection to identify risk factors.
• Developed and compared six machine learning models: Logistic regression, XGBoost, Random Forest, AdaBoost, Gradient Boosting Decision Tree, and KNN.
• Evaluated model performance using decision curves, calibration curves, and ROC metrics.

Main Results:

• Eleven risk factors were identified, including anemia, blood transfusion volume, blood loss, HCT, D-dimer, thrombin time (TT), CL, anesthesia duration, surgery duration, activated partial thromboplastin time (APTT), and postoperative pain scores.
• The Logistic regression model demonstrated the best performance and generalization ability.
• SHAP analysis revealed that shorter APTT, longer anesthesia duration, elevated TT, higher pain scores, reduced D-dimer, decreased CL, increased transfusion volume, increased blood loss, shorter surgery duration, and anemia are associated with increased DVT risk.

Conclusions:

• The Logistic model is optimal for predicting DVT risk after TKA.
• Identifying and managing identified risk factors can help reduce post-TKA DVT incidence.
• This predictive model provides a valuable reference for clinicians in evaluating and preventing DVT.