Measurement Inconsistencies in Consumer Smartwatch-Based Outdoor Gait Assessment: Implications for Real-World Monitoring in Parkinson’s Disease
Abstract
Zhenghua Li and Kenichi Yamamura
Background: We have sought to make objective gait assessment in Parkinson’s disease (PD) accessible in daily life using widely available consumer devices, rather than requiring specialized facilities or brief clinic-based tests. During repeated outdoor walking assessments, we encountered inconsistencies in distance and time records.
Objective: To characterize these inconsistencies and consider their implications for practical real-world assessment while recognizing genuine day-to-day motor fluctuations.
Methods: From August 16 to September 30, 2026, one participant with PD completed 22 distinct outdoor walking sessions in Harbin, China; one healthy adult completed two additional walks. An Apple Watch Ultra 2 and WorkOutDoors were used for recording; data were reviewed in HealthFit. Distance recorded by the device (recorded distance: RD) was compared with route distance measured using Google Maps (map distance: MD). The signed measurement error rate (MER) was defined as 100 × (RD − MD)/MD. Total time (TT), moving time (MT), and repeated imports were examined.
Results: Thirteen PD sessions were categorized as routine Tests and nine as Cases with conspicuous recording inconsistencies. One of two healthy-control walks also displayed a distance discrepancy, yielding 10 flagged sessions among 24 distinct walks (41.7%). Mean MER was +38.45 ± 19.92% for Tests and +28.28 ± 33.04% for Cases. Three stick-assisted walks showed negative MER (−21.2%, −15.7%, and −9.1%); the faster healthy-control walk showed −10.0%. In some records MT was nearly equal to TT, whereas in Case 4 MT exceeded TT by 1 min 44 s. Re-importing the same activity also altered displayed values.
Conclusion: Consumer-device gait assessment enables repeated real-world quantification without specialized facilities, but recorded distance and time require careful interpretation. Technical inconsistencies should be distinguished from genuine daily changes in PD walking ability. Simple review of route and time records can help preserve the accessibility of this approach.