What question did this study set out to answer?

This study aims to develop a machine learning framework for speech analysis to detect cognitive impairment and predict biomarkers related to Alzheimer’s disease.

May 8, 2026Open Access

Listening for Alzheimer’s clues: machine learning analysis of multidomain speech features for cognitive impairment screening

Key Points

This study aims to develop a machine learning framework for speech analysis to detect cognitive impairment and predict biomarkers related to Alzheimer’s disease.
Analyzed speech samples from 2,320 individuals across cognitive stages including subjective cognitive decline (SCD), mild cognitive impairment (MCI), and Alzheimer's dementia (ADD).
Utilized three spoken tasks to extract multidomain speech features such as acoustic, lexical, syntactic, and semantic traits.
Trained machine learning models to classify cognitive status and predict amyloid, tau, and neurodegeneration biomarkers (ATN).
Achieved AUC values of up to 0.94 for differentiating SCD vs. ADD and 0.82 for SCD vs. MCI classifications.
The models demonstrated AUCs of 0.71, 0.74, and 0.73 for ATN biomarker prediction.
Found strong correlations (up to 0.83) between speech-based metrics and cognitive function scores.

Abstract

Introduction Early detection of Alzheimer’s disease (AD) is critical for timely intervention, particularly during the mild cognitive impairment (MCI) stage. This study aimed to develop and evaluate a multidomain speech analysis framework to support cognitive screening, biomarker prediction within the amyloid, tau and neurodegeneration (ATN) framework, and estimation of cognitive function across the AD continuum. Methods This study analyzed speech from 2,320 individuals spanning the cognitive spectrum-including those with subjective cognitive decline (SCD), MCI, and Alzheimer’s disease dementia (ADD)-using three spoken tasks (∼3 min) and extracted multidomain features including acoustic, lexical, syntactic, and semantic features. Machine learning models were trained to classify cognitive status, predict amyloid, tau and neurodegeneration (ATN) biomarker positivity, and estimate scores across six neuropsychological domains. Results Multidomain speech models achieved high performance in differentiating cognitive stages, with AUC values of up to 0.94 for SCD vs. ADD and 0.82 for SCD vs. MCI classifications. In biomarker prediction, the models yielded AUCs of 0.71, 0.74, and 0.73 for ATN classification, respectively. Speech-based models also showed strong correlations (up to 0.83) with cognitive function scores. Feature importance analysis revealed that verbal fluency measures were the most predictive. Explainability analyses indicated minimal dependency on age, sex, or education, supporting model fairness. Discussion These findings show that multidomain speech features capture clinically and biologically relevant information across the AD continuum, enabling cognitive classification, biomarker prediction, and cognitive estimation. These results underscore the potential of speech analysis as a non-invasive, accessible tool for scalable cognitive screening and early detection of AD. These results underscore the potential of speech analysis as a non-invasive, accessible tool for scalable cognitive screening and early detection of AD.

اسأل الذكاء الاصطناعي

Bookmark

View Full Paper