Machine learning analysis of demographic, rehabilitation, and performance factors associated with Olympic medal productivity in elite athletes

Achieving an Olympic medal is widely regarded as the highest level of competitive sporting success. However, individual-level predictors associated with medal attainment have not been fully characterized. This study aimed to employ machine learning (ML) techniques to integrate diverse athlete data, including demographic characteristics, sports discipline, physiotherapy utilization, and movement performance measures, to predict medal outcomes at major international competitions. The analysis included 1,011 South Korean elite athletes who participated in the 19th Asian Games (Hangzhou; n = 867) and the 33rd Olympic Games (Paris; n = 144). Medal outcomes were harmonized across events using a common athlete-level framework that evaluated medal attainment irrespective of competition tier. The medal outcomes were evaluated across four targets: gold, silver, bronze, and overall medal counts. Gradient boosting algorithms (CatBoost, XGBoost, and LightGBM) were developed using nested cross-validation frameworks, and model interpretability was examined using SHapley Additive Explanations. XGBoost demonstrated the highest predictive performance for GMC, SMC, and BMC, with R 2 values of 0.447, 0.394, and 0.406, respectively, whereas CatBoost showed the highest performance for OMC prediction (R² = 0.590). SHAP analyses based on the representative XGBoost model indicated that sports discipline, total movement performance score, treatment received, and sex were among the most influential contributors across medal-count outcomes. Taken together, these findings indicate that stable athlete attributes, treatment-related variables, and movement performance measures were among the most influential predictors associated with medal outcomes within the proposed ML framework. The integration of athlete-specific data within interpretable ML frameworks may help inform individualized training design, rehabilitation planning, and future athlete monitoring.