Gut Microbiota Alterations Across REM Sleep Behavior Disorder and Parkinson's Disease: A Machine Learning-Based Meta-Analysis

Abstract

Recent studies have examined the relationship between rapid eye movement sleep behavior disorder (RBD), Parkinsons disease (PD), and the gut microbiota, but no consensus exists on the shared and distinct gut microbiota changes. This study aimed to identify consistent and divergent gut microbiota changes across RBD and PD and to evaluate the performance of machine learning (ML) models in distinguishing PD, RBD, and healthy controls (HC). A meta-analysis of four gut microbiota studies involving PD, RBD, and HC groups was conducted, comprising a total of 973 samples (379 PD, 251 RBD, and 343 HC). ML models could differentiate PD from HC (cross study validation (CSV) AUC 0.61 {+/-} 0.06) and RBD from HC (CSV AUC 0.58 {+/-} 0.03). However, distinguishing between PD and RBD was ineffective (CSV AUC 0.51 {+/-} 0.03). ML models distinguished PD and RBD from HC with weak to moderate predictive accuracy but failed to differentiate PD from RBD.