Enhancing Early Detection of Alzheimer's Disease: An Ensemble Model for Multi-Domain Cognitive Assessment Using Voice...

Area of Science:

• Neuroscience
• Biomarkers
• Digital Health

Background:

• Early Alzheimer's disease (AD) detection is critical but faces challenges with invasive and costly traditional methods.
• There is a significant need for accessible, non-invasive biomarkers for precise AD screening.
• Cognitive impairments in AD affect domains like language, memory, and attention.

Purpose of the Study:

• To develop and validate a multi-modal digital biomarker framework for accurate, non-invasive Alzheimer's disease detection.
• To assess the framework's performance across multiple cognitive domains using voice and video data.
• To enhance screening accessibility and real-world applicability.

Main Methods:

• A multi-modal digital biomarker framework was proposed, analyzing voice and video data.
• The framework evaluated cognitive functions including language, working memory, and visuospatial attention.
• An ensemble strategy was employed to integrate results from individual cognitive tasks.

Main Results:

• Individual cognitive tasks showed F1-scores from 69.23% to 77.78% and sensitivities from 69.23% to 80.77%.
• The ensemble strategy significantly improved detection performance.
• The final framework achieved an F1-score of 83.64% and a sensitivity of 88.46% in a dataset of 128 participants (51 AD, 77 HCs).

Conclusions:

• The proposed multi-modal digital biomarker framework offers a highly accurate solution for non-invasive Alzheimer's disease screening.
• Ensemble methods enhance the framework's detection capabilities, improving scalability and practicality.
• This digital approach represents a significant advancement in accessible and early AD detection.