A Speech-Based Mobile Screening Tool for Mild Cognitive Impairment: Technical Performance and User Engagement...

Area of Science:

• Gerontology
• Computational Linguistics
• Biomedical Engineering

Background:

• Traditional Mild Cognitive Impairment (MCI) screening methods lack accessibility and scalability.
• Developing innovative, user-friendly tools is crucial for early detection of cognitive decline.
• Speech-based assessments offer a promising avenue for remote and widespread cognitive screening.

Purpose of the Study:

• To develop and validate a speech-based automatic screening application for Mild Cognitive Impairment (MCI).
• To evaluate the app's performance against manual assessments and in real-world conditions.
• To assess user engagement, technology acceptance, and factors influencing perceived task benefit.

Main Methods:

• Development of a speech-based app with three speech-language tasks, user-centered design, and server-client architecture.
• Integration of automated speech processing and Support Vector Machine (SVM) classifiers for MCI detection.
• Validation through comparison with manual assessment (n=12) and real-world testing (n=22), including user engagement surveys.

Main Results:

• The app demonstrated comparable performance to manual assessment (F1 = 0.93 vs. 0.95) and maintained reliability in real-world settings (F1 = 0.86).
• User engagement analysis showed high technology acceptance (86%) and a significant link between cognitive exercise habits and perceived task benefit (p < 0.01).
• Task difficulty did not correlate with cognitive performance (p = 0.119), indicating broad accessibility.

Conclusions:

• The mobile application shows robust assessment capabilities and sustained user engagement for MCI screening.
• This speech-based approach has potential for widespread cognitive screening in the geriatric population.
• The user-centered design and technology integration facilitate accessible and scalable cognitive health monitoring.