Technology and Dementia Preconference.

Area of Science:

• Digital health
• Cognitive neuroscience
• Global health equity

Background:

• Digital speech assessments show promise for early detection of cognitive decline and dementia.
• Existing tools primarily serve English-speaking, high-income countries, creating global health disparities.
• The Toolkit to Examine Lifelike Language (TELL) was developed to address these inequities by focusing on low- and middle-income countries.

Purpose of the Study:

• To outline the adaptations and challenges in localizing the TELL app for Latin American and African countries, specifically Chile and Kenya.
• To refine digital speech testing tools for diverse linguistic, cultural, and technological contexts.

Main Methods:

• Interdisciplinary teams collaborated to customize TELL's interface, prompts, data acquisition, preprocessing, metrics, and visualizations.
• App texts were translated into Spanish and Kiswahili using AI and human review.
• Speech elicitation stimuli were selected via culture-specific and culture-neutral pathways, with enhanced data collection for videoconferencing and offline modes.

Main Results:

• A five-step audio preprocessing system was implemented to handle varied acquisition environments.
• Tailored metrics for Spanish and Kiswahili were developed using language-agnostic, language-specific, cross-linguistic, and multilingual resources.
• User feedback guided optimizations for clinical metric visualizations, highlighting successes and areas for improvement.

Conclusions:

• Cross-cultural adaptation of speech testing technologies presents significant hurdles, solutions, and compromises.
• Lessons learned offer best practices for optimizing digital assessments in underserved global regions.
• This work contributes to more equitable access to cognitive health technologies worldwide.