Researchers of a single-center study (n=101) used machine learning and network analysis to identify factors shaping health-related quality of life (HRQoL; Parkinson’s disease questionnaire [PDQ]-39) in Parkinson’s disease. Across models, anxiety, fear of falling (FES-K), Hoehn and Yahr stage, treatment duration, and wearable-derived gait irregularity (higher sample entropy of acceleration/gyroscope during forward walking and fast turning; lower/irregular “max jerk”) were the strongest correlates of worse overall HRQoL. Domain-specific signals echoed this pattern: mobility was linked to FES-K, unified Parkinson’s disease rating scale (UPDRS) total, freezing of gait, shorter 6-minute walk test (6MWT) distance, and gait entropy; activities of daily living to UPDRS, FES-K, self-efficacy for exercise, short physical performance battery, and entropy metrics; and emotional well-being to anxiety, depression, turning gyroscope metrics, and 6MWT. Network modeling placed fear of falling and anxiety near the PDQ-39 hub, with connections to gait entropy measures, highlighting interdependence between psychological factors and sensor-based motor instability.
Clinically, the findings suggest that combining psychological screening (anxiety, fear of falling) with objective digital gait biomarkers can better stratify HRQoL risk and guide targeted interventions (eg, balance/gait training under challenging conditions like fast walking/turns, fall-confidence programs, aerobic capacity building, and mood treatment). The authors propose larger, multiethnic, longitudinal studies—ideally leveraging home-based, longer-duration wearable recordings—to validate sensor-informed pathways and refine decision-support tools for Parkinson’s disease care.
Reference: Hwang J, Youm C, Park H,. Multidimensional factors of health-related quality of life in parkinson’s disease using ensemble learning and network analysis. Sci Rep. 2025;15(1):36786. doi: 10.1038/s41598-025-20656-9.
