guest ::
| Home > Publications database > Machine learning to diagnose, classify and predict phenoconversion in isolated REM sleep behavior disorder |
| Home > Publications database > Machine learning to diagnose, classify and predict phenoconversion in isolated REM sleep behavior disorder |
| Journal Article | FZJ-2026-02580 |
; ; ; ; ; ; ; ; ; ; ; ; ;
2026
Elsevier
Kidlington, Oxford [u.a.]
This record in other databases:
Please use a persistent id in citations: doi:10.1016/j.smrv.2026.102298 doi:10.34734/FZJ-2026-02580
Abstract: Rapid eye movement (REM) sleep behaviour disorder (RBD), particularly its idiopathic/isolated form (iRBD), is a prodromal marker for α-synucleinopathies, including Parkinson's disease, dementia with Lewy bodies and multiple system atrophy. Machine learning (ML) offers opportunities to improve diagnosis and risk stratification in this high-risk group. We conducted a systematic review of PubMed, Embase (Ovid) and Medline (Ovid) from 2014 to September 2025, following PRISMA guidelines. From 335 records identified, 202 remained after duplicate removal and 75 studies on adult humans with clinically diagnosed RBD or iRBD that applied and validated an ML model were included. Fifty-eight studies addressed diagnosis, four studied RBD phenotypes, and thirteen evaluated prediction of phenoconversion to overt α-synucleinopathy. Across diagnostic studies, reported accuracies ranged from ∼63% to ∼99.7%, with median values around 90%, using polysomnography, EEG, neuroimaging, molecular and behavioural markers. Phenoconversion models (often using dopaminergic imaging or multimodal features) achieved AUCs up to ∼0.94, but frequently relied on small, single-centre cohorts with heterogeneous definitions of phenoconversion and limited external validation. A wide variety of ML algorithms was used (n ~ 30), most commonly support vector machines, random forests and logistic regression. Overall, ML approaches show promise for scalable diagnosis and risk stratification in iRBD, but progress is constrained by methodological bias, inconsistent endpoints, data imbalance and a lack of explainable, externally validated models. We outline methodological priorities to make future ML tools clinically interpretable and translatable.Keywords: Dementia with Lewy bodies; Machine learning; Multiple system atrophy; Parkinson's disease; REM sleep behaviour disorder; α-synucleinopathies.
; 
; 
; Clarivate Analytics Master Journal List ; Essential Science Indicators ; IF >= 10 ; JCR ; SCOPUS ; Science Citation Index Expanded ; Web of Science Core Collection
|
The record appears in these collections: |
PDF