Voice Assistants for Health Self-Management: Designing for and with Older Adults

TL;DR

This paper tackles two core challenges in health self-management for older adults—health awareness and adherence to medical regimens—by designing and evaluating a personal health assistant powered by an LLM integrated with Alexa. Through a five-stage process (in-home interviews, initial prototyping, co-design workshops, prototype refinement, and in-home validation), the authors create a context-aware VA capable of debriefing after-visit summaries and generating tailored medication reminders. The in-home validation shows high usability (mean SUS of 85) and successful handling of debriefing, health queries, and reminder creation, emphasizing design priorities such as personalization, user context, and autonomy. The work advances VAs for health self-management with practical insights on user-centered design, safety considerations, and adaptable interaction patterns, while also outlining limitations and avenues for future longitudinal studies and broader populations.

Abstract

Supporting older adults in health self-management is crucial for promoting independent aging, particularly given the growing strain on healthcare systems. While voice assistants (VAs) hold the potential to support aging in place, they often lack tailored assistance and present usability challenges. We addressed these issues through a five-stage design process with older adults to develop a personal health assistant. Starting with in-home interviews (N = 17), we identified two primary challenges in older adult's health self-management: health awareness and medical adherence. To address these challenges, we developed a high-fidelity LLM-powered VA prototype to debrief doctor's after-visit summary and generate tailored medication reminders. We refined our prototype with feedback from co-design workshops (N = 10) and validated its usability through in-home studies (N = 5). Our work highlights key design features for personal health assistants and provides broader insights into desirable VA characteristics, including personalization, adapting to user context, and respect for user autonomy.

Figures (5)

• Figure 1: We designed personal health voice assistant to support older adults' self-management in 5-stages: stage 1---interviews with older adults; stage 2---initial prototyping; stage 3---co-design workshops with older adults; stage 4---refining the VA prototype; and stage 5---in-home validation study with older adults. All activities with participants were approved by our institutional review board.
• Figure 2: We present the interview findings alongside their underlying causes. Based on these findings, we outline the design requirements and demonstrate their implementation in our initial prototype.
• Figure 3: Findings of co-design workshop for consultation, reminder and tracking. Our prototype does not focus on tracking.
• Figure 4: Co-design workshop findings: general considerations for VA design. The pink circles represent considerations that were implemented in the refined prototype's phase 2 and 3 through prompt engineering.
• Figure 5: Refined VA and example participant interaction https://youtu.be/i2-Ewn7_Rjw. "…" represents truncated speech for brevity. C1--C4 are conversation labels used in the text to reference examples in this figure. All prompts in supplementary materials$^1$.