Accelerometry-Derived REM Sleep Behavior Disorder Score Is Associated with Incident Parkinson's Disease in the UK Biobank
- 1Icahn School of Medicine at Mount Sinai
Abstract
Estimating Parkinson's disease (PD) risk years before diagnosis remains an unmet need. We applied a validated machine learning classifier for REM sleep behavior disorder (RBD) detection to 7-day wrist accelerometry data in 87,975 UK Biobank participants followed for 10 years. Participants in the highest RBD risk stratum (>99th percentile) had an approximately fivefold increased hazard of incident PD compared with the lowest-risk group (0-90th percentile), with a dose-dependent relationship across the full score distribution. Among non-converters, higher RBD-classifier scores were associated with baseline cognitive deficits and greater burden of autonomic and psychiatric prodromal features. The association with incident PD remained independent of PD polygenic risk score, while RBD-classifier score and genetic risk were synergistic. The combined high-risk group achieved a positive likelihood ratio of 7.91, approximately threefold higher than questionnaire-based RBD screening. At this High-risk threshold over the 10-year horizon, the classifier showed a sensitivity of 7.4%, specificity of 99.2%, positive predictive value of 7.5%, and negative predictive value of 99.2% for incident PD (N=50,816). These findings support wrist accelerometry as a scalable, high-specificity approach for prodromal PD risk enrichment.
Related
- Gut microbiota and short-chain fatty acids in idiopathic REM sleep behavior disorder and Parkinson’s disease
- Impaired Dynamic Cerebral Autoregulation Across the REM Sleep Behavior Disorder Spectrum: Evidence From Isolated REM Sleep Without Atonia to Parkinson's Disease With RBD.
- Depressive symptoms increase during follow-up and are associated with phenoconversion in isolated REM sleep behavior disorder
- Probable Rapid Eye Movement Sleep Behavior Disorder Associated With Escitalopram in a Patient With Schizoaffective Disorder: A Case Report.
- Accelerometer-derived 24-hour behavior composition and incident clinically diagnosed sleep apnoea: a prospective cohort study
BibTeX
@misc{mejia2026accelerometryderived,
title = {{Accelerometry-Derived REM Sleep Behavior Disorder Score Is Associated with Incident Parkinson's Disease in the UK Biobank}},
author = {G. R. Mejia and A. Brink-Kjaer and L. Liu and L. Zhou and K. Gunter and K. H. Ryu and S. D. Wickramaratne and A. Parekh and Z. Gan-Or and E. During},
howpublished = {medRxiv},
year = {2026},
month = jul,
doi = {10.64898/2026.07.02.26356952},
url = {https://www.medrxiv.org/content/10.64898/2026.07.02.26356952v2},
note = {CC BY},
}
RIS
TY - UNPB
TI - Accelerometry-Derived REM Sleep Behavior Disorder Score Is Associated with Incident Parkinson's Disease in the UK Biobank
AU - G. R. Mejia
AU - A. Brink-Kjaer
AU - L. Liu
AU - L. Zhou
AU - K. Gunter
AU - K. H. Ryu
AU - S. D. Wickramaratne
AU - A. Parekh
AU - Z. Gan-Or
AU - E. During
PY - 2026
DA - 2026/07/02
PB - medRxiv
DO - 10.64898/2026.07.02.26356952
UR - https://www.medrxiv.org/content/10.64898/2026.07.02.26356952v2
AB - Estimating Parkinson's disease (PD) risk years before diagnosis remains an unmet need. We applied a validated machine learning classifier for REM sleep behavior disorder (RBD) detection to 7-day wrist accelerometry data in 87,975 UK Biobank participants followed for 10 years. Participants in the highest RBD risk stratum (>99th percentile) had an approximately fivefold increased hazard of incident PD compared with the lowest-risk group (0-90th percentile), with a dose-dependent relationship across the full score distribution. Among non-converters, higher RBD-classifier scores were associated with baseline cognitive deficits and greater burden of autonomic and psychiatric prodromal features. The association with incident PD remained independent of PD polygenic risk score, while RBD-classifier score and genetic risk were synergistic. The combined high-risk group achieved a positive likelihood ratio of 7.91, approximately threefold higher than questionnaire-based RBD screening. At this High-risk threshold over the 10-year horizon, the classifier showed a sensitivity of 7.4%, specificity of 99.2%, positive predictive value of 7.5%, and negative predictive value of 99.2% for incident PD (N=50,816). These findings support wrist accelerometry as a scalable, high-specificity approach for prodromal PD risk enrichment.
ER -