Predicting ACL Reconstruction Failure with Machine Learning: Development of Machine Learning Prediction Models.

Area of Science:

• Orthopedic Surgery
• Biomedical Data Science
• Machine Learning in Medicine

Background:

• Anterior cruciate ligament reconstruction (ACLR) is standard for ACL injuries, but failure remains a challenge.
• Existing statistical models for predicting ACLR failure often have suboptimal predictive efficacy.
• There is a need for improved methods to identify patients at risk of ACLR failure.

Purpose of the Study:

• To evaluate the predictive performance of various machine learning algorithms for ACLR failure.
• To identify the most significant predictors associated with ACLR failure.
• To enhance the accuracy of predicting outcomes after ACLR.

Main Methods:

• A cohort of 680 patients undergoing ACLR was analyzed.
• Nine machine learning algorithms were trained and validated on routinely collected data.
• Model performance was assessed using the area under the receiver operating characteristic curve (AUC).

Main Results:

• Machine learning models demonstrated good predictive performance, with AUCs ranging from 0.71 to 0.85.
• The CatBoost classifier (AUC 0.85) and random forest classifier (AUC 0.84) showed the highest predictive accuracy.
• Knee hyperextension was consistently identified as the primary predictor of ACLR failure across all models.

Conclusions:

• Machine learning algorithms are effective tools for predicting ACLR failure.
• Knee hyperextension is a significant and consistent risk factor for ACLR failure.
• These findings support the integration of machine learning into clinical decision-making for ACLR.