Perceived social support in the daily life of people with Parkinson's disease: a distinct role and potential classifier.

Motor outcomes in Parkinson's disease (PD) have long been the primary diagnostic criteria and treatment targets. While non-motor outcomes of PD impact daily well-being, they are rarely targeted by interventions or utilized for classification. Despite promising evidence, the contributions of perceived social support (PSS) to PD detection and well-being in real-world settings remain unclear. Using remote monitoring technologies, we investigated the relationship between PSS and three non-motor measures-cognition, anxiety, and depression-in 92 participants: 45 PD and 47 matched-controls. To examine the specificity of PSS to non-motor features, we also examined the associations between PSS and three motor-related measures: disease severity, duration, or stage. Moreover, we developed machine-learning classifiers (ML) based on only non-motor features to identify disease status (PD/controls) in two cohorts: low and high PSS. PSS was significantly associated with non-motor measures in PD, with stronger correlations than in matched-controls in real-world settings. However, no significant correlations were found between PSS and the three motor-related measures, demonstrating PSS's limitations. While the ML classification models performed low in high-PSS, they classified 13% better in a low-PSS cohort (AUC = 0.8), demonstrating moderate-high discriminatory performance. Taken together, our findings underscore the role of PSS in PD, highlighting its distinct contributions to non-motor classification models and the daily well-being of patients.