To understand how patients make sense of AI-drafted online messaging technologies in patient portals and to identify implications for patient-centered implementation.
Approach:
Study Design: Conducted a qualitative study using in-depth, semistructured interviews with adult patients from a large academic health system.
Participant Recruitment: Used purposive sampling to recruit a diverse group of respondents, focusing on those with varying demographic characteristics and prior survey responses.
Data Collection: Interviews lasted approximately 45 to 60 minutes and included discussion of standardized vignette scenarios related to AI involvement in messaging.
Data Analysis: Employed an abductive analytic approach with a multidisciplinary team to develop and refine themes from the interview data.
Key Findings:
Patients generally preferred AI-drafted responses over clinician-written responses but rated satisfaction lower when AI involvement was disclosed.
Patient evaluations are influenced by both message content and beliefs about AI's role in clinical relationships.
Unresolved questions remain regarding acceptable AI use, efficiency versus relational expectations, and norms of disclosure.
Interpretation:
Limitations:
The study's findings may not be generalizable beyond the specific patient population sampled.
Potential bias in participant selection, as those with lower satisfaction were specifically targeted.
by Kellie Owens, Athmeya Jayaram, Anand Chowdhury, Kathryn I. Pollak, Sam Klotman, Zachary Griffen, Miles Danielski, Ben Goldstein, Jennifer Maddocks, Matthew Roman, Eric G. Poon, Armando Bedoya, Joanna Cavalier
Investigational inhibitor was not associated with treatment-related serious adverse events and produced biomarker changes consistent with pathway inhibition in healthy volunteers.
