An interpretable, clinically grounded framework for digital speech biomarker development in neurodegenerative...

Area of Science:

• Neuroscience
• Speech Science
• Artificial Intelligence

Background:

• Communication disorders are common in neurodegenerative diseases, impacting quality of life.
• Current management requires personalized interventions for progressive communication impairments.
• Objective tools for assessing speech changes and variability are needed.

Purpose of the Study:

• To develop an AI-based acoustic framework for automated speech marker extraction.
• To enable objective assessment and phenotyping of communication disorders in neurodegenerative diseases.
• To support personalized, measurement-based care.

Main Methods:

• Analyzed speech samples from individuals with ALS, Parkinson's disease, and healthy controls.
• Extracted 50 acoustic features, clustered into six composite markers.
• Evaluated markers for associations with functional communication, disease differentiation, and phenotyping.

Main Results:

• Markers detected subclinical speech changes before significant functional decline.
• Successfully differentiated disease-specific communication impairments (AUC > 0.90).
• Identified distinct speech subgroups within each disease.

Conclusions:

• The AI framework shows potential for clinical translation.
• Enables early detection, differential diagnosis, and phenotyping of communication disorders.
• Advances personalized care for neurodegenerative diseases.