Supplementary Material for: Machine Learning for Establishing the Precision Prediction of Sarcopenia

As the global population ages, sarcopenia is increasingly recognized for its severe consequences, including disability, falls, injuries, hospitalization, and even death. Despite its significance, research on predicting sarcopenia using machine learning is limited. This study aims to develop an effective machine learning model for sarcopenia prediction. Data from 1,441 participants were retrospectively reviewed from Kaohsiung Medical University-affiliated hospitals between 2022 and 2024. The dataset included demographics, lifestyle habits, medical history, and other relevant factors, with sarcopenia assessment based on the Asian Working Group for Sarcopenia (AWGS) 2019 criteria. Six machine learning models were evaluated: CatBoost, K Nearest Neighbor (KNN), Naive Bayes (NB), Random Forest (RF), Gradient Boosting Decision Tree (GBDT), and Extreme Gradient Boosting (XGBoost). Model performance was assessed using accuracy, precision, recall, and F1-Score. Feature importance and SHAP (Shapley Additive Explanations) were used for feature analysis. CatBoost outperformed the other models, achieving an accuracy of 96.62%, with similarly high precision, recall, and F1-Score. Feature importance analysis using SHAP revealed that age, gender, pulse rate, pulmonary disease, blood pressure, dizziness, and missing teeth were key predictors in the model for sarcopenia prediction. The findings suggest that the CatBoost model is a highly effective tool for predicting sarcopenia, offering potential for early detection and intervention.