Early Diagnosis of Parkinson's Disease Through Voice Analysis and Machine Learning

Abstract

Parkinson’s Disease (PD) is a progressive neurological disorder, and its prevalence has increased considerably in recent years. Early and accurate diagnosis plays a crucial role in effective treatment, but identifying the disease in its early stages remains challenging, especially for non-specialist clinicians. This study presents a predictive analytics framework for detecting Parkinson's disease using voice-based features. To improve the diversity and reliability of the data, datasets from UCI and Kaggle were combined. The data were preprocessed, and important vocal features such as jitter, shimmer, and pitch were extracted using Praat. Several machine learning models, including KNN, SVM, Random Forest, Logistic Regression, and ANN, were applied and evaluated independently. Among these, Logistic Regression showed the best performance, achieving an accuracy of 98.3% and a sensitivity of 97%. The results indicate that combining datasets can improve model performance and reliability. Overall, this study highlights the potential of using multiple datasets and machine learning models for more reliable Parkinson’s disease prediction.