Machine-learning-assisted CRISPR/Cas12a biosensing for biomarker identification of neurodegenerative movement disorders

Neurodegenerative movement disorders (NMDs) such as Parkinson’s disease (PD), progressive supranuclear palsy (PSP) and multiple system atrophy (MSA) share overlapping clinical features yet distinct pathology, complicating early diagnosis. We present a machine-learning-assisted, glass-fiber-based CRISPR-Cas12a immunosensing platform (GUSCI) for multiplexed quantification of five extracellular-vesicle (EV) biomarkers in paired serum and serum-derived CD81⁺ EVs. Across 340 clinical samples, integrated EV biomarker profiles supported exploratory classification analyses and generally showed stronger discriminatory signals than matched serum markers. EV-derived 4R Tau enabled discrimination of PSP (AUC = 0.848 (95% CI 0.7379–0.9362)), whereas combined EV signatures of IL-6, TNF-α and α-synuclein distinguished PD (AUC = 0.823 (95% CI 0.7185–0.8988)). These findings support serum-derived EV biomarker profiling by GUSCI as a promising exploratory framework for biomarker discovery and differential diagnosis in NMDs.