Utilizing speech analysis to differentiate progressive supranuclear palsy from Parkinson's disease

Area of Science:

• Neurology
• Speech Science
• Digital Health Technology

Background:

• Differentiating early Parkinson's disease (PD) from Progressive Supranuclear Palsy (PSP) is crucial for accurate patient management and clinical trial stratification.
• Early-stage diagnostic accuracy impacts treatment efficacy and patient prognosis.

Purpose of the Study:

• To investigate the utility of a digital health platform in distinguishing between PD and PSP based on speech characteristics.
• To assess the feasibility of using speech biomarkers for early differential diagnosis.

Main Methods:

• Twenty-one participants (11 PSP, 10 PD) underwent standardized speech tasks (reading, counting, phonation) recorded via a tablet.
• Speech features were analyzed using specialized software, with statistical comparisons and machine learning classification employed.
• Correlations between speech measures and clinical scores (e.g., PSPRS, MoCA) were examined.

Main Results:

• Significant differences in articulation rate, speech-to-pause ratio, and pause duration were observed during passage reading between PSP and PD groups.
• Phonation analysis revealed distinct spectral characteristics for PSP patients.
• Machine learning models achieved high accuracy (AUC 0.93, sensitivity 0.95, specificity 0.90) in classifying PSP and PD based on speech features.

Conclusions:

• Digital health technology offers a feasible method for differentiating early-stage PSP from PD through speech analysis.
• Speech biomarkers hold promise for non-invasive differential diagnosis of neurodegenerative movement disorders.
• Further multi-center validation is recommended to confirm these findings.