Machine Learning Model Using Speech Acoustics Identifies Neurodegenerative Diseases With High Accuracy

A recent study in the IEEE Transactions on Neural Systems and Rehabilitation Engineering highlights the accuracy of speech markers in distinguishing neurodegenerative diseases. Using acoustic properties alone, the model achieved an 82% accuracy rate in identifying Friedreich ataxia (FA), multiple sclerosis (MS), and healthy controls. The 21 identified acoustic features, categorized as spectral qualia, spectral power, and speech rate, serve as robust markers for these diseases.

Key Takeaways:
– Speech markers, analyzed through machine learning, show high accuracy in diagnosing neurodegenerative diseases.
– Identified acoustic features provide potential diagnostic promise.
– Machine learning and speech analysis offer opportunities for initial detection, disease monitoring, and refined test selection.

The study suggests the potential of multiclass supervised machine learning to discriminate between diseases, indicating a shift beyond traditional healthy-pathological distinctions. Leveraging big data for speech analysis across diverse patient populations could enhance disease detection. This innovative approach opens possibilities for healthcare in remote areas where access to providers is limited, offering a valuable tool for initial diagnosis and ongoing monitoring.


Source NeurologyLive

Author: Neurologica