Benchmarking speech biomarkers of Alzheimer's against cognitive and neural measures - PubMed
3 hours ago
- #Speech Biomarkers
- #Alzheimer's Disease
- #Neuroimaging
- The study aimed to validate digital speech biomarkers (DSBs) for detecting Alzheimer's disease (AD) in Latino populations by comparing them to traditional cognitive tests and neuroimaging measures.
- Participants included 33 AD patients and 33 healthy controls who underwent verbal fluency tasks, episodic memory and executive function tests, MRI for brain volume, and fMRI for connectivity.
- Machine learning classifiers using DSBs derived from verbal fluency performed comparably (AUC = 0.84) to those based on episodic memory (AUC = 0.90), executive function (AUC = 0.79), and structural MRI (AUC = 0.90), and outperformed functional MRI-based classifiers (AUC = 0.65).
- Key DSB features, such as word length and frequency, were top discriminators and correlated with right (pre)frontal brain volume after adjusting for sociodemographic factors.