The Analysis of Digital Biomarkers for Identifying and Predicting Frailty and Healthy Elderly through Machine Learning

This study explores using an AI model to identify and predict frailty in elderly individuals by analyzing digital biomarkers of head and hand movements during virtual reality (VR) tasks. A sample of 20 participants, divided equally into healthy and frail groups, was assessed using traditional questionnaires and VR-based cognitive activities. The study utilized machine learning (ML) algorithms, including Decision Tree, Random Forest, and Logistic Regression, to analyze the digital biomarkers and predict the health status of the subjects. The results demonstrated that the Logistic Regression model achieved an accuracy of 0.83 and a ROC-AUC of 0.83, indicating its reliability in classifying frailty and healthy elderly. The research highlights the potential of combining digital biomarkers and VR with ML techniques to detect frailty conditions, suggesting a novel assessment modality that could enhance early interventions and improve the quality of life for the elderly.