Feature Selection and Machine Learning Approaches for Detecting Sarcopenia Through Predictive Modeling
Akhrorbek Tukhtaev, Dilmurod Turimov, Jiyoun Kim, Wooseong Kim
Abstract
Sarcopenia is an age-associated condition characterized by a muscle mass and function decline. This condition poses significant health risks for the elderly. This study developed a machine-learning model to predict sarcopenia using data from 664 participants. Key features were identified using the Local Interpretable Model-Agnostic Explanations (LIME) method. This enhanced model interpretability. Additionally, the CatBoost algorithm was used for training, and SMOTE-Tomek addressed dataset imbalance. Notably, the reduced-feature model outperformed the full-feature model, achieving an accuracy of 0.89 and an AUC of 0.94. The results highlight the importance of feature selection for improving model efficiency and interpretability in clinical applications. This approach provides valuable insights into the early identification and management of sarcopenia, contributing to better patient outcomes.