Machine learning approaches to identify Parkinson's disease using voice signal features

Area of Science:

• Neurology
• Computer Science
• Artificial Intelligence

Background:

• Parkinson's Disease (PD) is a common neurodegenerative disorder with challenging diagnosis due to overlapping symptoms with aging and essential tremor.
• Early detection is crucial for managing PD progression and maintaining patient lifestyles, despite the absence of a cure.

Purpose of the Study:

• To develop and evaluate Machine Learning (ML) and Deep Learning (DL) models for accurate Parkinson's Disease detection using voice signal features.
• To differentiate between healthy individuals and PD patients through voice analysis.

Main Methods:

• Utilized voice recordings from 31 PD patients and healthy individuals (195 recordings total) from the UCI Machine Learning Repository.
• Applied various ML/DL models including Support Vector Machine (SVM), Random Forest (RF), Decision Tree (DT), K-Nearest Neighbor (KNN), and Multi-Layer Perceptron (MLP).
• Employed techniques like Synthetic Minority Over-sampling Technique (SMOTE), Feature Selection, and hyperparameter tuning (GridSearchCV) to optimize model performance.

Main Results:

• The Multi-Layer Perceptron (MLP) model achieved 98.31% accuracy, 98% recall, 100% precision, and 99% F1-score.
• The Support Vector Machine (SVM) model demonstrated 95% accuracy, 96% recall, 98% precision, and 97% F1-score.
• Both MLP and SVM models, optimized with GridSearchCV and SMOTE, showed superior performance in distinguishing PD patients from healthy individuals.

Conclusions:

• The proposed ML/DL models, particularly MLP and SVM, reliably predict Parkinson's Disease using voice features.
• The findings suggest that this AI-driven approach can be integrated into healthcare systems for efficient and accurate PD diagnosis.
• Voice analysis presents a promising, non-invasive method for early PD detection and management.