Reliability of a Home-Based Smartphone Balance Assessment in Healthy Middle-Aged and Older Adults

Smartphone accelerometry could enable longitudinal home-based balance testing, yet its reliability across tasks and outcome measures must be established. We assessed the test-retest reliability and measurement precision of a custom smartphone balance application. Sixty-nine healthy, community-dwelling middle-aged and older adults (ages 41-76 years) performed a 5 s home-based balance assessment weekly for 3 weeks. The assessment was directed by the application, which generated accelerometer-based sway metrics. Participants performed two 30 s trials each of feet together and tandem stance (eyes open, eyes closed) and single leg stance (eyes open only). Intraclass correlation coefficients (ICCs) and relative standard error of measurement (SEM%) were calculated for time-domain and frequency-domain measures. Outcomes achieved good-to-excellent reliability when averaged across three weekly assessments (ICC = 0.59-0.95), particularly mediolaterally. Time-domain measures, particularly mean acceleration (ICC = 0.84-0.92; SEM% = 11.1-23.5%), were the most reliable and precise outcomes overall, while frequency-domain measures showed lower reliability and precision at higher spectral bands (ICC as low as 0.59; SEM% up to 41.5%). As standing tasks became more difficult, reliability and precision declined correspondingly. We conclude that minimally supervised, home-based balance assessment can produce reliable, precise sway measures in healthy, screened, iPhone-owning middle-aged and older adults. This application carries potential for large-scale, remote balance monitoring for research purposes, though validation in more diverse and higher-risk populations will be needed before clinical application can be recommended.