Development of Machine Learning Models with Explainable AI for Frailty Risk Prediction and Their Web-Based Application in Community Public Health

Seungmi Kim¹Byung Kwan Choi²Jae Il Lee²Up Huh^3,4,5Myung-Jun Shin^3,6,7Zoran Obradovic^8,9Daniel J. Rubin¹⁰Jong-Hwan Park^1,11,12*

• ¹Pusan National University, Busan, Republic of Korea
• ²Department of Neurosurgery, Medical Research Institute, Pusan, Republic of Korea
• ³Biomedical Research Institute, Pusan National University Hospital, Busan, Republic of Korea
• ⁴Department of Thoracic and Cardiovascular Surgery, Pusan National University School of Medicine, Yangsan, Republic of Korea
• ⁵Department of Thoracic and Cardiovascular Surgery, Pusan National University Hospital, Busan, Republic of Korea
• ⁶Department of Rehabilitation Medicine, Pusan National University School of Medicine, Yangsan, Republic of Korea
• ⁷Department of Rehabilitation Medicine, Pusan National University Yangsan Hospital, Yangsan, Republic of Korea
• ⁸Temple University Center for Data Analytics and Biomedical Informatics, Philadelphia, United States
• ⁹Department of Computer and Information Sciences, Temple University, Philadelphia, PA, United States
• ¹⁰Temple University Lewis Katz School of Medicine, Philadelphia, United States
• ¹¹Department of Clinical Bio-Convergence, Graduate School of Convergence in Biomedical Science, Yangsan, Republic of Korea
• ¹²Convergence Medical Institute of Technology, Pusan National University Hospital, Busan, Republic of Korea

Background: Frailty is a public health concern linked to falls, disability, and mortality. Early screening and tailored interventions can mitigate adverse outcomes, but community settings require tools that are both accurate and explainable. Korea is entering a super-aged phase, yet few approaches have used nationally representative survey data. Objective: This study aimed to identify key predictors of frailty risk based on the K-FRAIL scale using explainable machine learning (ML) with data from the 2023 National Survey of Older Koreans, and to develop and internally validate prediction models. In addition, to demonstrate the potential applicability of the developed models in both community public health and clinical practice, a web-based application was implemented. Methods: Data from 10,078 older adults were analyzed, with frailty defined by the K-FRAIL scale (robust = 0, pre-frail = 1–2, frail = 3–5). A total of 132 candidate variables were constructed through selection and derivation. Using CatBoost with out-of-fold SHAP (SHapley Additive exPlanations, a game-theoretic approach to quantify feature contributions), 15 key predictors were identified and applied across ten algorithms under nested cross-validation. Model performance was evaluated using ROC-AUC, PR-AUC, F1-score, balanced accuracy, and the Brier score. To assess feasibility, a single-page bilingual web application integrating the CatBoost inference pipeline for offline use was developed. Results: SHAP analysis identified depression score, age, IADL count, sleep quality, and cognition as leading predictors, followed by smartphone use, number of medications, province, driving status, hospital use, physical activity, osteoporosis, eating alone, digital adaptation difficulty, and sex, yielding 15 key predictors across mental, functional, lifestyle, social, and digital domains. Using these predictors, boosting models outperformed other algorithms, with CatBoost achieving the best performance (ROC-AUC = 0.813 ± 0.014; PR-AUC = 0.748 ± 0.019). Conclusion: An explainable machine-learning model with strong discrimination and adequate calibration was developed, and a lightweight web application was implemented for potential use in community and clinical settings. However, external validation, recalibration, and subgroup fairness assessments are needed to ensure generalizability and clinical adoption.