Conversational Agents in Behavioral Sleep Medicine: Designing Self-Report and Analytics Tools

TL;DR

The paper addresses the limitations of traditional sleep diaries in Behavioral Sleep Medicine by prototyping a large-language-model–driven, voice-based sleep diary and a specialist-facing analytics tool. Through a three-stage design—interviews with sleep specialists, iterative prototyping with design workshops, and focus groups—the authors derive design requirements and validate a multi-panel CA analytics interface that integrates both subjective diary data and objective metrics. Findings reveal potential to reduce patient and clinician burden, capture richer contextual information, and extend care beyond the clinic, while also highlighting risks related to LLM reliability and data privacy. The work offers concrete design implications, including cross-platform diary interactions, integrated data visualizations, AI-assisted analysis, and safe, clinician-guided patient engagement, with implications for at-home care and broader behavioral health contexts.

Abstract

The sleep diary is a widely used clinical tool for understanding and treating sleep disorders in Behavioral Sleep Medicine (BSM); however, low patient compliance and limited capture of contextual information constrain its effectiveness and leave specialists with an incomplete picture of patients' sleep-related behaviors. In this work, we explore conversational agents (CAs) as an alternative to traditional diary methods by designing a voice-based sleep diary and a specialist-facing analytics tool, and using them as design probes to understand how CAs might support BSM more broadly. Our multi-stage study with specialists comprised: (1) interviews to identify shortcomings of current text-based diaries, (2) iterative co-design of the conversational diary and analytics tool, and (3) focus groups to examine broader opportunities for CAs in BSM. This work offers empirical insights into how specialists envision CAs in clinical care and outlines design implications for integrating them into existing self-report practices and behavioral interventions.

Figures (7)

• Figure 1: Our research process unfolded in three stages. In Stage 1, we interviewed behavioral sleep specialists to identify shortcomings of their current practice with text-based sleep diaries. In Stage 2, we designed a voice assistant sleep diary and conducted a 4-week deployment study with university students (proxy for patients) to collect data, followed by the iterative development of a specialist-facing analytics tool through initial prototyping, co-design workshops, prototype refinement, and a validation study. In Stage 3, we conducted focus groups with specialists to explore the broader potential of conversational agents in BSM.
• Figure 2: Interview findings: Behavioral sleep medicine workflow and shortcomings in the use of traditional sleep diaries. We identified three key challenges in BSM: (1) patient burden, (2) lack of context reporting, and (3) specialist burden. Based on specialists' vision for an ideal sleep diary tool, we outline design requirements for our prototype (Stage 2). The part 1, part 2 and part 3 of the prototype (stage 2) are color coded throughout the paper based on the design requirements.
• Figure 3: Stage 2: LLM-powered VA sleep diary four-week data collection study with 15 university students as proxy patients. The raw diary conversations are then converted to structured JSON objects for use in the analytics tool.
• Figure 4: Stage 2: Initial prototype for LLM-powered specialist-facing interface (part 2 and 3). Qualitative and quantitative insights along with report generations are implemented in the five panels: Summary, Visualization, Conversation, Chat Bubble, and Report.
• Figure 5: Stage 2: Co-design workshop findings for specialist-facing interface. To refine the interface, various features under three themes---data visualization, AI-assisted analysis, and report generation---are implemented in the five panels: Summary, Visualization, Conversation, Chat Bubble, and Report.