Parkinson’s Disease Detection based on Voice Analysis

Parkinson’s Disease (PD) is a gradually crippling neurodegenerative condition that mostly affects motor activities. PD symptoms are referred to as parkinsonism. The patient's quality of life is greatly impacted by these symptoms, which include bradykinesia, stiffness, tremors, and abnormalities in gait. These symptoms appear gradually. The illness is frequently accompanied by non-motor symptoms like anxiety, depression, and behavioural abnormalities in addition to cognitive impairment. The novel application of voice analysis in PD detection is the main emphasis of this study. The goal is to create a machine learning model that uses vocal traits to reliably diagnose PD. A comprehensive dataset was curated; encompassing voice samples from both PD affected individuals and healthy controls. Machine learning algorithms were applied, utilizing 60% of the data for training and 40% for testing. The model achieved a detection efficiency of 73.8%, underscoring the potential of voice-based diagnostics for PD. The model was trained using a dataset consisting of 24 columns, each of which represented a symptom value. The 'status' column indicated whether PD was present (1) or absent (0). The findings imply that voice analysis has potential as a non-invasive, preliminary PD diagnostic method. Further refinements and integration with medical assessment could enhance its clinical utility, contributing to improved patient care and timely interventions.