Voice-Based Detection of Parkinson's Disease Using Machine and Deep Learning Approaches: A Systematic Review.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Speech Science

Background:

• Parkinson's disease (PD) is a progressive neurodegenerative disorder with early vocal impairments.
• Voice analysis with ML and DL offers a non-invasive approach for early PD detection.

Purpose of the Study:

• To systematically review recent advancements in ML and DL for voice-based PD detection.
• To identify current challenges and future directions in the field.

Main Methods:

• Systematic literature review of studies from 2020-2025 across major databases.
• Analysis of 69 studies focusing on datasets, speech tasks, feature extraction, models, and validation.
• Comparative assessment of classical ML and DL model performance.

Main Results:

• Classical ML models (SVMs, RFs) performed well on small datasets.
• Deep learning (CNNs, RNNs, Transformers) showed greater robustness and scalability.
• Challenges include dataset heterogeneity, class imbalance, and inconsistent validation.

Conclusions:

• The field is shifting towards self-supervised learning for better generalizability.
• Future progress requires large, multilingual datasets, standardized protocols, and interpretable AI for clinical translation.