Wearable Devices Show Selective Benefits in Parkinson’s Disease
Key Takeaways
- In randomized wearable-device interventions for people with Parkinson disease, significant pooled improvement was reported for walking speed, Timed Up and Go test performance, and UPDRS-III scores.
- Most other pooled outcomes were reported as not significantly different between groups, although reporting for step cadence was internally inconsistent.
- Walking-speed benefit appeared in subgroup analyses of longer programs and midrange weekly training frequency, while shorter courses and other frequency categories were not significant.
- Walking-speed effects did not separate significantly by device category or by feedback modality.
- The authors cautioned that the evidence base was limited by modest trial sizes, short intervention and follow-up periods, heterogeneous protocols, and publication-bias assessment that was largely limited to walking speed.
A systematic review and meta-analysis of wearable devices in Parkinson disease pooled randomized controlled trials (RCTs) in people with Parkinson disease, reported under Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and prospectively registered in PROSPERO (CRD420251246681). Searches of international and Chinese databases through December 2025 identified 13 RCTs with 380 participants from Germany, Italy, Canada, South Korea, Japan, Brazil, and Belgium, comparing wearable-device training with training without wearable devices or with no training across gait, balance, motor, endurance, and quality-of-life outcomes.
Included-study ages ranged from 60 to 76 years, intervention duration from 1 to 12 weeks, and training frequency from 1 to 7 times per week. Outcomes included the Freezing of Gait Questionnaire (FOGQ), Berg Balance Scale (BBS), Mini-Balance Evaluation Systems Test (Mini-BESTest), Timed Up and Go test (TUGT), double-support (DS) time, Unified Parkinson’s Disease Rating Scale, Part III (UPDRS-III), Six-Minute Walk Test (6MWT), and Parkinson’s Disease Questionnaire (PDQ). Three crossover trials were summarized separately because published paired data could not be integrated into the primary pooling.
Walking speed improved in 9 studies with 300 participants (MD 0.07, 95% CI 0.01 to 0.12, p=0.02). TUGT performance improved in 4 studies with 141 participants (MD -2.00, 95% CI -3.57 to -0.43, p=0.01). UPDRS-III scores improved in 7 studies with 228 participants (SMD -0.33, 95% CI -0.59 to -0.06, p=0.02).
The authors overall reported no significant between-group differences for most other pooled outcomes, including stride length, step length, step cadence, FOGQ, BBS, Mini-BESTest, DS time, 6MWT, and PDQ. For walking speed, subgroup signals were reported for interventions lasting 7 to 12 weeks (MD 0.14, 95% CI 0.04 to 0.23, p=0.01) and for schedules of 3 to 5 times per week (MD 0.07, 95% CI 0.00 to 0.14, p=0.04), while shorter duration, other weekly-frequency categories, device category, and feedback modality were not significant modifiers. A supplementary scale-specific analysis qualitatively favored UPDRS-III over the Movement Disorder Society-sponsored revision, and publication-bias testing was reported only for walking speed, without substantial evidence of bias.
The authors described wearable devices as a heterogeneous umbrella category rather than a single intervention mechanism, spanning different cueing and feedback approaches. They also noted modest sample sizes, variability in device types and intervention protocols, short intervention and follow-up periods, and the difficulty of blinding visible rehabilitation devices. Because most outcomes were informed by few trials, formal publication-bias testing was not available for most endpoints, and the duration and frequency subgroup signals should be viewed as preliminary.
Overall, the review associated wearable-device interventions in Parkinson disease with selective gains in walking speed, dynamic balance, and motor scores, while broader effects on gait parameters, balance scales, endurance, and quality of life were not established across pooled endpoints.
Clinician Questions
What counted as a wearable-device intervention in the Parkinson disease trials?
The intervention category included wearable-device training compared with training without wearable devices or with no training, and it spanned cueing and feedback approaches using auditory, proprioceptive, combined visual-auditory, and vision-proprioception feedback. The authors treated these devices as a heterogeneous umbrella category rather than as a single mechanism.
Why were crossover wearable-device trials not included in the primary Parkinson disease meta-analysis?
Three crossover trials were summarized separately because the published reports did not provide enough paired data to reconstruct within-participant treatment effects for pooling with the parallel-group trials. Their overall direction was described as generally consistent with the main analysis, but they were not incorporated into the primary estimate.
Which patients were represented in the randomized wearable-device trials for Parkinson disease?
The review included people diagnosed with Parkinson disease without restrictions by sex, nationality, ethnicity, or Hoehn and Yahr stage. In practice, the evidence came from mostly older adults enrolled in multinational trials across Europe, Asia, Canada, and Brazil.
Why was publication-bias testing reported only for walking speed in this Parkinson disease review?
Most pooled outcomes were informed by too few studies for formal publication-bias assessment, so the authors evaluated publication bias only for walking speed. They did not identify substantial evidence of bias for that endpoint.