Objective: To evaluate the diagnostic performance of artificial intelligence (AI)-based voice analysis in detecting Parkinson’s disease.

Background: Speech impairment is an early feature of PD. Recent advances in artificial intelligence have enabled the analysis of subtle acoustic abnormalities that may serve as digital biomarkers for early disease detection.

Method: A systematic search of PubMed, Scopus and IEEE databases was performed to identify studies assessing AI-based voice analysis for PD detection. Studies evaluating machine learning algorithms using acoustic voice features were included. Diagnostic accuracy measures were extracted and summarized.

Results: Fourteen studies met the inclusion criteria. Several machine learning algorithms, including support vector machines and deep neural networks, demonstrated diagnostic accuracies ranging from 78% to 92% in differentiating individuals with PD from healthy controls. Key predictive features included vocal tremor, reduced pitch variability and phonation instability.

Conclusion: AI-based voice analysis shows promising diagnostic potential for early PD detection. These digital biomarkers could enable scalable screening approaches, particularly in resource-limited settings where access to neurological evaluation is limited.

« Back to 2026 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/artificial-intelligence-based-voice-analysis-for-early-detection-of-parkinsons-disease-a-systematic-review/