Transforming Alzheimer's Digital Caregiving through Large Language Models

Area of Science:

• Gerontology
• Artificial Intelligence
• Digital Health

Background:

• Alzheimer's Disease and Related Dementias (AD/ADRD) impose significant burdens on informal caregivers.
• Existing caregiving education materials (CEMs) and mobile application descriptions (MADs) offer a foundation for digital interventions.
• The need for adaptive and evolving support systems for AD/ADRD caregivers is increasingly recognized.

Purpose of the Study:

• To investigate the potential of AI-driven Large Language Models (LLMs) in developing digital caregiving strategies for AD/ADRD.
• To analyze existing CEMs and MADs to identify key caregiving tasks and digital functionalities.
• To align caregiving tasks with digital solutions across different stages of AD progression.

Main Methods:

• Analysis of 38 CEMs from MedlinePlus and 57 AD digital caregiving MADs.
• Utilized ChatGPT 3.5 to extract essential caregiving tasks and match them with appropriate digital functionalities for each AD stage.
• Identified digital literacy requirements for caregivers supporting individuals with AD/ADRD.

Main Results:

• AD caregiving was categorized into four stages (Pre-Clinical, Mild, Moderate, Severe) with associated key tasks.
• Digital aids identified include memory-enhancing apps, GPS tracking, and voice-controlled devices.
• Six essential digital literacy skills were identified: basic digital skills, communication, information management, safety/privacy, healthcare knowledge, and caregiver coordination.

Conclusions:

• Advocates for an LLM-driven strategy to design digital caregiving interventions for AD/ADRD.
• Proposes a novel paradigm for AD/ADRD support offering adaptive assistance tailored to caregiver needs.
• Aims to enhance caregiver shared decision-making and patient care capabilities through AI-powered digital tools.