Sophia Johnson
Research Mentor: Nirali Shah
Mentor Department: Physical Medicine and Rehabilitation, Medicine
Author(s): Cristina Daraban, Sophia Johnson, Nirali Shah
Session: Session 1 (9:00 AM – 9:50 AM)
Presentation Type: Poster 85
Abstract
Objective. Despite growing interest in artificial intelligence-based tools for chronic illness management, little is known about how patients with rare autoimmune diseases engage with these technologies. This study analyzes the differences in chat content topics in individuals with Systemic Sclerosis (SSc) who have high vs low engagement while interacting with an AI-health coach. Methods. We conducted a mixed-method study where 20 individuals with SSc interacted with an AI (LLM-driven) health coach over a period of 4 weeks. We used chat duration (active seconds), user messages, and session numbers to create a composite z-score for each participant. Participants with z-scores >0 were classified as high engagement. Conversely, participants with =0 were classified as low engagement. For qualitative analysis, we conducted content analysis outlined by Theish et al. Of the 20 transcripts, 4 were coded together among 4 researchers. Once agreeability was achieved, the remaining transcripts were coded individually by 2 researchers. Data integration will be conducted post-analysis of qualitative data to understand differences in content topics by engagement levels. Results. Twelve individuals with SSc had z-scores = 0, indicating low engagement, and 8 individuals had z-scores >0, indicating high engagement. We are currently analyzing qualitative participant data. Conclusion. The findings of this study will help us understand how individuals with SSc interact with an AI health coach. Insights into patient needs are guiding refinements to our AI health coach. We are enhancing personalization while integrating medical referral prompts to account for the AI’s inability to provide professional healthcare. Overall, our research provides a framework to help us understand how we can refine management needs for other rheumatic diseases and further deliberate amongst researchers on the impact of AI.
