Development of a Frailty Detection Model Using Machine Learning with the Korean Frailty and Aging Cohort Study Data.

Area of Science:

• Gerontology
• Computational Medicine
• Biostatistics

Background:

• Frailty is a significant health concern in the elderly population, impacting morbidity and mortality.
• Existing frailty assessment criteria can be enhanced through advanced analytical methods.
• Identifying novel predictive factors is crucial for early intervention and improved geriatric care.

Purpose of the Study:

• To leverage machine learning for discovering new predictive factors of frailty in older adults.
• To validate the efficacy of machine learning models in identifying frailty indicators.
• To build upon established frailty criteria using computational approaches.

Main Methods:

• Utilized data from the Korean Frailty and Aging Cohort Study (KFACS), classifying 1,066 robust and 165 frail participants.
• Employed feature selection techniques to identify 27 meaningful frailty predictors.
• Trained and validated models using Support Vector Machine, Random Forest, and Gradient Boosting algorithms with stratified 10-fold cross-validation and hyperparameter optimization.

Main Results:

• Identified 27 significant features predictive of frailty.
• Achieved a weighted average F1-score of 95.30% using the Random Forest algorithm.
• Demonstrated the model's reliability on an unseen test set.

Conclusions:

• Machine learning can effectively augment existing frailty assessment criteria.
• The developed method provides rapid analysis of questionnaire data for large-scale health assessments.
• This approach can proactively alert individuals and healthcare providers to potential frailty risks.