Quantitative Evaluation of Freezing of Gait in Parkinson’s Disease Using Smartwatch Telemetry
Abstract
Zhenghua Li and Kenichi Yamamura
Background: Freezing of Gait (FOG) and daily motor fluctuations severely impair independence in Parkinson’s disease (PD). Continuous monitoring via consumer smartwatches offers a non-invasive way to measure real-world Digital Mobility Outcomes (DMOs). However, FOG-induced standstill and environmental variations cause measurement biases that are poorly characterized.
Objective: To quantitatively evaluate Freezing of Gait (FOG) episodes in Parkinson’s disease using smartwatch telemetry, establish a simple and clinically intuitive Range Ratio metric (Mean ± Range Ratio %) to capture true day-to-day symptom variability without relying on standard deviation, assess FOG-induced GPS trajectory distance distortion, and investigate the potential moderating effects of ambient biometeorological factors (temperature and relative humidity) on gait motor performance.
Methods: Field gait trials were conducted along an outdoor circuit (Fukuoka Park) ranging from 240 m to 800 m comparing Control sessions (n=4) and Test sessions exhibiting FOG (n=15). High-frequency time-series sensor data were recorded using the WorkOutDoors application on an Apple Watch Ultra 2 and analyzed via HealthFit and custom Excel processing pipelines. Day-to-day symptom variability across the 15 Test sessions was expressed using the Range Ratio (Mean±Range Ratio %), defined as [(Max−Min)/Mean]×100.
Results: High-precision telemetry successfully extracted continuous spatiotemporal metrics. Test sessions exhibited significantly higher FOG Ratio (0.320 vs. 0.037, p = 2.26 × 10^-7) and massive GPS distance inflation (32.10% vs. 1.25%, p = 7.94 × 10^-8) compared to Controls. FOG Ratio showed strong inverse correlations with walking speed (r = -0.742) and step length (r = -0.685), confirming that spatial gait breakdown closely mirrors freezing severity. Across 15 test sessions, day-to-day symptom ranges expressed as Mean ± Range Ratio % were: FOG Ratio 0.32 ± 145.6%, Walking Speed 3.12 ± 55.8% km/h, Cadence 125.6 ± 39.9% spm, and Step Length 45.8 ± 27.1% cm. Ambient temperature showed no significant association with FOG Ratio (r = 0.188, p = 0.503) or GPS error (r = 0.284, p = 0.305). Conversely, relative humidity demonstrated a statistically significant inverse correlation with FOG Ratio (r = -0.594, p = 0.020) and GPS error ratio (r = -0.656, p = 0.008).
Conclusion: FOG Ratio derived from smartwatch telemetry is an exceptionally sensitive biomarker for capturing real- world day-to-day motor symptom swings (145.6% fluctuation). GPS distance inflation during FOG motor arrest (~32%) mandates the integration of inertial zero-velocity filters in DMO algorithms. Environmental humidity modulates freezing severity, presenting new biometeorological insights for PD motor management.