PARKINSONS DISEASE DETECTION USING MACHINE LEARNING

Parkinson’s disease (PD) is a progressive neurodegenerative disorder that affects more than ten million people worldwide. Early detection remains a major clinical challenge because motor symptoms typically emerge after a significant proportion of dopaminergic neurons have already been lost. Voice impairment, however, is one of the earliest non-motor manifestations of PD, making acoustic biomarker analysis an attractive, non-invasive screening modality. This paper presents NeuroDetect AI, a fully client-side web application that screens for Parkinson’s disease using thirteen voice biomarkers derived from the Oxford Parkinson’s Telemonitoring Dataset. The system implements a logistic regression classifier in TypeScript, exposes a clean React-based interface for both manual entry and CSV upload, and produces an interpretable verdict accompanied by a probability, confidence score and risk level (Low / Moderate / High). The application is designed for use by clinicians, researchers and informed patients as a triage tool that complements—but does not replace—clinical evaluation. Experimental validation on synthetic samples derived from the Oxford dataset yields screening accuracy comparable to published baselines while running entirely in the browser, requiring no server-side computation and preserving patient privacy by design.