Journal of Sleep Research

Movement-based model distinguished Parkinson’s disease

The best model identified 80.5% of Parkinson’s cases in a comparison with controls who had restless legs syndrome, using five days of recordings.

Our read

Biomarkers
Studied:
Case-control study of patients with PD and controls, against controls with restless legs syndrome
What changed:
The best model identified 80.5% of Parkinson’s cases in a comparison with controls who had restless legs syndrome
The catch:
The isolated sleep-disorder group included only six men

Editor of record: Boris Djordjevic · AgingResearch.News

Follow @agingbrief

Researchers retrospectively analyzed movement recordings from 42 people with Parkinson’s disease without rapid eye movement (REM) sleep behavior disorder, 45 with both disorders and 47 controls with restless legs syndrome. They also tested six men with isolated REM sleep behavior disorder, which involves acting out dreams. Mean ages were 62.0, 65.2, 61.4 and 65.3 years, respectively; the first three groups included 21, 28 and 35 men.

In this observational study, researchers compared artificial intelligence models using five days of recordings per person. The best model identified 80.5% of Parkinson’s cases. It also distinguished Parkinson’s with and without the sleep disorder, identifying 84.4% of those with it and correctly classifying 92.9% of those without it. The six men with the isolated sleep disorder had scores between the comparison groups for both tasks.

Why it matters

Isolated REM sleep behavior disorder can precede Parkinson’s disease. The authors see movement-based models as potential tools for identifying early movement changes and eventually predicting progression.

Caveats

This retrospective analysis evaluated classification within the same cohort, not prediction of future Parkinson’s disease; validation in groups followed over time remains necessary. Controls had restless legs syndrome, and the isolated sleep-disorder group included only six men.

Comments

Be the first to comment.