JMIR Publications

ABSTRACT

Background:

Artificial intelligence–based clinical decision support systems (AI-CDSS) are increasingly integrated into medical practice, creating hybrid decision-making processes in which physicians and AI systems jointly contribute to clinical judgments. While these systems may improve diagnostic accuracy and efficiency, their impact on patients’ trust in and adherence to hybrid medical decisions remains insufficiently understood.

Objective:

This study examined how the type and the timing of physician AI support influence potential patients’ trust in the medical decisions, perceptions of the hybrid decision-making process, and intentions to follow the medical advice.

Methods:

Two preregistered vignette-based online experiments were conducted with members of the general public in Germany. In Study 1 (N=489), participants imagined four medical consultations, in which the physician used no AI support, descriptive AI support (informational/visual assistance), or diagnostic AI support (preliminary diagnostic suggestions). Study 2 (N=570) additionally manipulated whether the physician reviewed AI advice after having made an independent own assessment (sequential decision-making) or not (concurrent decision-making). Participants rated their trust in the medical decisions, trustworthiness of the medical provider, uniqueness neglect, and willingness to follow the medical advice on 7-point Likert scales, with greater values representing stronger agreement. Linear mixed-effects models were used for quantitative analyses. Recurring themes were extracted from open-ended responses.

Results:

In Study 1, the physician’s use of diagnostic AI support compared to descriptive AI support produced significantly lower mean ratings of trust in the medical decisions (5.00 vs. 5.37, t486=3.51, P=.002) and perceived provider trustworthiness (4.93 vs. 5.40, t(486)=4.46, P<.001) as well as higher mean ratings of perceived uniqueness neglect (3.24 vs. 2.93, t(486)=2.72, P=.021), but no significant differences regarding the willingness to follow the advice (5.42 vs 5.62, t(486)=1.74, P=.247). Study 2 revealed a significant interaction effect between AI Support Type and AI Support Timing for trust in the medical decisions (t(436)=2.71, P=.007), perceived provider trustworthiness (t(436)=2.78, P=.006), and perceived uniqueness (t(436)=-2.34, P=.020), but not for willingness to follow the advice (t(436)=1.81, P=.070). Specifically, diagnostic AI support was evaluated less favorably than descriptive AI support when the physician reviewed primary medical information and AI output simultaneously, but not when the physician first assessed the primary medical information independently before reviewing the AI output. Qualitative analyses of open-ended responses identified strategies to build patient trust, addressing concerns about incorrect AI results influencing physicians’ judgments.

Conclusions:

Our findings suggest that trust in AI-supported medical decisions depends not only on whether a physician uses AI support but more so on subjective perceptions of how the AI support is used. By underscoring the importance of transparent communication and thoughtful physician-AI interaction, our mixed-methods findings contribute to ongoing discussions on how to design and implement AI-CDSS that are not only effective but also accepted by their human stakeholders.