Ensemble Learning for Parkinson's Disease Diagnosis: A Comparative Study for different Speech Features

A form of neurodegenerative disorder known as Parkinson's disease affects millions of individuals globally and manifests with a variety of motors and non-motor signs and symptoms. When neurons that release dopamine in certain parts of the brain malfunction or die, people have trouble in accomplishing basic activities. The degree of seriousness on the patients' condition increases as a result of those signs becoming worse with time. This approach predicts parkinson's disease that uses a novel XGBoost ensemble method based on signals generated by speech. To ensure data quality and relevance, an initial processing step was taken on recordings of speech from individuals with and without Parkinson's disease. Various speech characteristics, such as the pitch change, nervousness, glitter, MFCCs, and statistical metrics, were collected in order to capture the crucial components of speech that are linked to parkinson's disease.