Related Experiment Video
Updated: Aug 24, 2025
10:28
Dynamic Digital Biomarkers of Motor and Cognitive Function in Parkinson's Disease
Published on: July 24, 2019
15.3K
Cognitive Digital Biomarkers from Automated Transcription of Spoken Language
N Tavabi1, D Stück, A Signorini
1Rhoda Au, 72 E. Concord Street, Boston University School of Medicine, Boston, MA 02118. Telephone: (617) 358-0089;
The Journal of Prevention of Alzheimer'S Disease
|October 25, 2022
Summary
Voice analysis can predict cognitive decline in neurodegenerative diseases. Combining acoustic and linguistic features from speech significantly improves early dementia detection accuracy.
Area of Science:
- Neurology
- Computational Linguistics
- Biomarkers
Background:
- Early detection of Alzheimer's disease and other neurodegenerative disorders is crucial but challenging.
- Digital voice recording and analysis of spoken responses offer a potential avenue for predicting cognitive status.
- Existing diagnostic methods lack reliability for early-stage cognitive impairment.
Purpose of the Study:
- To investigate the predictive power of voice recordings from neuropsychological testing for cognitive status.
- To determine if acoustic and linguistic features in speech can serve as digital biomarkers for cognitive impairment.
- To assess the feasibility of using automated voice analysis for early dementia detection.
Main Methods:
- Compared acoustic and paralinguistic variables from automated transcriptions with linguistic variables from manual transcriptions.
- Trained a logistic regression classifier to predict cognitive status using voice data.
- Tested the classifier's accuracy against actual diagnoses in a cohort study.
Main Results:
- Demographic features alone showed limited predictive power (AUROC 0.60).
- Incorporating clinical and acoustic features improved prediction (AUROC 0.81).
- Adding transcription-related linguistic features further enhanced predictive accuracy (AUROC 0.90).
Conclusions:
- Voice-based digital biomarkers, processed automatically, can aid in early dementia detection.
- Combining voice analysis with standard screening methods offers a scalable approach.
- Speech analysis holds promise as a non-invasive tool for cognitive health assessment.
