Assessing gait using wearable digital health technologies in Spinal Bulbar Muscular Atrophy: an analytical validation study


Topic:

Translational Research

Poster Number: P389

Author(s):

Kruthika Doreswamy, MPH, National Institutes of Health Clinical Center, Rakesh Pilkar, PhD, ActiGraph, L.L.C., Milan Barnes, MPT, National Institutes of Health, Michaele Sheehan, MSPT, National Institutes of Health, Angela Kokkinis, RN, National Institutes of Health, Abdullah AlQahtani, M.D., M.P.H., National Institutes of Health, Galen Joe, MD, National Institutes of Health, Christopher Grunseich, MD, National Institutes of Health, Christine Guo, PhD, ActiGraph, L.L.C., Minal Jain, PT, DSc, FAPTA, National Institutes of Health

Background: Spinal bulbar muscular atrophy (SBMA) is a rare, X-linked inherited neuromuscular disease. SBMA affects males, and symptoms begin with hand tremors and lower proximal muscle weakness, leading to impaired mobility, physical function, and poor quality of life. The 6-minute walk test (6MWT) is a commonly used clinical tool to assess functional gait, aerobic capacity, and endurance. Wearable digital health technologies (DHTs) can objectively record high-frequency movement data in clinic as well as at home.

Objectives: This study aims to evaluate the validity of DHT-derived measures of gait during the 6MWT in individuals with SBMA.

Methods: Twelve males participated (age: 60 ± 11.5 years, height: 1.79 ± 0.07 m, weight: 95.96 ± 21.16 kg). Five GT9X (ActiGraph, L.L.C., Pensacola, FL) sensors were placed on wrists, ankles, and lumbar spine. Algorithms specific to sensor location were applied to derive step count, cadence, distance and gait speed. For the ankle location, stride velocity 95th percentile (SV95C) was also calculated. The 6MWT was videotaped and two raters annotated the steps to provide the reference data for step count and cadence. Total distance covered was measured in meters. Estimated measures were compared against the reference data to calculate mean absolute percent error (MAPE) and agreement metrics (Pearson’s r, Bland-Altman statistics). MAPE <15% is considered as acceptable accuracy for outcome measures. Results: Sensor-derived gait measures strongly correlated with the reference values (p<0.05) for all sensor locations. For the step count, ankle (MAPE<14%) and lumbar (MAPE <9%)-derived measures showed higher accuracies. Ankle-derived gait speed and SV95C provided the highest accuracies (MAPE<6%) compared to lumbar (MAPE<15%) and wrist (MAPE<34%). The distance measure showed higher MAPE across the three sensor locations. Conclusions: Although limited by a small sample size and heterogeneity among study participants, the current evidence supports the feasibility and validity of DHTs-derived gait measures in SBMA.