Idiopathic rapid eye movement sleep behavior disorder (iRBD) is recognized as a key prodromal stage of neurodegenerative α-synucleinopathies, such as Parkinson’s disease (PD) and dementia with Lewy bodies (DLB). Individuals with iRBD have a heightened risk of developing these conditions over time, following two primary phenoconversion trajectories: Parkinsonism-first and dementia-first.
Early identification of non-invasive biomarkers that can predict which trajectory an individual will follow remains an important clinical challenge. Wearable sensor technology holds promise in this domain by enabling objective gait analysis, which may reveal subtle motor impairments that precede clinical diagnosis.
Prior work has established that gait abnormalities are detectable even before phenoconversion in people with iRBD, but the predictive utility of these abnormalities remains unclear. Traditional instrumented gait assessment systems provide precise measurements but are limited by cost and complexity.
On the other hand, wearable sensors, which are portable and cost-effective, offer the potential for scalable and real-world longitudinal monitoring. Previous studies have yielded mixed findings, possibly due to methodological differences and heterogeneity in iRBD cohorts. Additionally, the timing of phenoconversion varies widely, complicating cross-sectional interpretations.
The study at hand addresses this gap by applying a multi-sensor protocol in a well-characterized cohort of iRBD patients, investigating associations with both parkinsonism-first and dementia-first phenoconversion paths.
The Current Study
Participants underwent comprehensive baseline assessments including detailed neurological examinations and cognitive testing to establish baseline clinical status. All gait evaluations were conducted under standardized laboratory conditions to ensure consistency across participants and time points.
The study enrolled 68 patients diagnosed with iRBD through video-polysomnography according to accepted diagnostic criteria and 61 healthy controls matched for age. All participants underwent comprehensive neurological assessments to exclude other conditions.
The cohort was followed longitudinally for an average of 3.68 years, during which 21 participants phenoconverted to neurodegenerative disease, primarily PD or DLB. Gait was quantitatively assessed using six inertial sensors (sampling at 128 Hz) placed bilaterally on wrists and ankles, as well as on the sternum and lumbar regions.
Participants performed one-minute walking trials under three conditions: normal pace, fast walking, and a cognitively demanding dual-task condition involving serial subtraction. Key gait features extracted included stride length, arm swing range and velocity, gait variability, and asymmetry, calculated using a specialized analytics platform.
Dual-task cost was also derived to quantify gait changes under cognitive load. Statistical analyses, adjusted for age, sex, and BMI, employed Cox proportional hazards models, competing risks regressions, and generalized estimating equations to assess associations between baseline and longitudinal gait parameters and phenoconversion status and clinical trajectories.
Gait Signals Parkinson’s Risk
At baseline, individuals with iRBD demonstrated significantly altered gait characteristics relative to controls, including reduced stride length, increased variability, and diminished arm swing. Notably, specific gait features - including shortened stride length, increased swing time variability, and reduced arm swing range and peak velocity - showed stronger associations with conversion to Parkinson’s disease rather than DLB.
Among those who converted, higher asymmetry in peak arm swing velocity was observed compared with non-converters and dementia-first converters, suggesting early lateralized motor dysfunction linked to parkinsonism. Longitudinal data revealed a more rapid decline in stride length under dual-task conditions and a greater increase in dual-task cost for converters, indicating progressive impairment in gait automaticity and dual-tasking ability.
These changes may reflect underlying dopaminergic and cholinergic deficits contributing to motor and cognitive symptoms. Importantly, gait parameters were more predictive for Parkinsonism-first phenoconversion, while no significant gait predictors emerged for dementia-first conversion, highlighting phenotypic heterogeneity in disease progression.
The study emphasizes the unique sensitivity of wearable sensor-based gait metrics to early motor impairments in iRBD and supports their utility as digital biomarkers for disease monitoring. Comparison with previous research underscores the advantages of multi-sensor configurations for capturing both lower-limb and arm movement parameters, which are especially relevant given that upper-limb motor asymmetry often manifests earlier in Parkinson’s disease.
Early Detection Using Wearables
This investigation provides strong evidence that wearable sensor-derived gait analysis can serve as an effective, non-invasive tool for predicting phenoconversion and delineating clinical trajectories among individuals with idiopathic REM sleep behavior disorder. The use of multiple inertial sensors enables comprehensive characterization of gait and arm movements, capturing subtle motor changes that precede clinical diagnosis, particularly along the Parkinsonism-first route.
Additionally, gait alterations under dual-task paradigms highlight cognitive-motor interplay and progression-sensitive metrics suitable for longitudinal monitoring. While acknowledging limitations such as sample size and the need for multicenter validation, the study advocates integrating wearable sensor technology into clinical practice and future therapeutic trials targeting prodromal α-synucleinopathies. This approach has the potential to enhance early risk stratification, facilitate targeted interventions, and improve outcome assessments through objective digital biomarkers.
Journal Reference
Cen S., Zhang H., et al. (2026). Association of wearable sensor-based gait analysis with phenoconversion trajectories in idiopathic REM sleep behavior disorder. npj Parkinsons Disease. DOI: 10.1038/s41531-026-01334-7, https://www.nature.com/articles/s41531-026-01334-7