Summary

Our study finds that machine learning algorithms may be used to compare the risk of failure of specific patient-procedure combinations in the treatment of cartilage defects of the knee.

Abstract

Background

Many treatment options exist for focal cartilage defects of the knee. There is, however, a lack of evidence-based methods to determine the optimal treatment of these injuries.

Purpose

To develop machine learning algorithms to predict failure of surgical procedures that address cartilage defects of the knee and detect the most valuable variables associated with failure.
Study Design: Case-control; Level of evidence, 3.

Methods

A single institution prospectively collected database of cartilage procedures was queried for procedures performed between 2000 and 2018. Failure was defined as revision cartilage surgery and/or knee arthroplasty. One hundred and one preoperative and intraoperative features were evaluated as potential predictors. The dataset was randomly divided into training (70%) and independent testing (30%) sets. Four machine learning algorithms were trained and internally validated. Algorithm performance was assessed using area under curve (AUC) and the Brier score. Local Interpretable Model-agnostic Explanations (LIME) was utilized to assess the optimized algorithm fidelity.

Results

A total of 1091 patients who underwent surgical procedures addressing cartilage defects in the knee with a minimum of 2-years of follow-up were included. The most-common procedure was chondroplasty (n=560; 51%) followed by osteochondral allograft transplantation (n=306; 28%), microfracture (n=150; 14%), autologous chondrocyte implantation (n=39; 4%), and osteochondral autograft transplantation (n=36; 3%). The mean follow-up was 3.5±2.8 years. The mean age was 40.5±15 years. There were 205 (18.8%) patients who failed at final follow-up. The Random Forest algorithm was found to be the best performing algorithm, with an AUC of 0.765 and a Brier score of 0.135. The most important features for predicting failure following surgical procedures addressing cartilage defects of the knee were symptom duration, age, body mass index (BMI), and lesion grade. LIME analysis provided a patient-specific comparison for the risk of failure of an individual patient being assigned various types of cartilage procedures.

Conclusion

Machine learning algorithms were accurate in predicting the risk of failure following cartilage procedures of the knee, with the most important features in descending order being symptom duration, age, BMI, and lesion grade. Integrated human and machine learning decision-making may improve patient selection and bring about the new era of patient-tailored evidence-based clinical care.