JMIR Publications

ABSTRACT

Background:

Artificial intelligence (AI) chatbots and conversational tools are increasingly used by patients to seek health information outside clinical encounters. However, patients may simultaneously perceive these tools as useful while expressing concerns about accuracy, trust, privacy, and the role of health care professionals. Understanding how these considerations combine within distinct patient viewpoints is important for the responsible implementation of patient-facing AI.

Objective:

This study aimed to identify and characterize shared viewpoints among patients in Saudi Arabia regarding the use of AI tools for seeking health information.

Methods:

A Q-methodology study was conducted, informed by the unified theory of acceptance and use of technology (UTAUT). A literature-derived concourse was refined to a final Q set of 40 statements. A purposively recruited P set of 24 patients from the western region of Saudi Arabia was structured to provide variation in age, gender, disease status, and place of residence. Participants completed the study entirely online using Easy HTMLQ, ranking the 40 statements on a forced quasi-normal distribution ranging from −5 (most strongly disagree) to +5 (most strongly agree), followed by post-sort questions explaining key placements. By-person factor analysis was performed using KADE software, with centroid factor extraction and varimax rotation. Factor retention was guided by multiple statistical criteria and substantive interpretability, and post-sort responses were used to contextualize factor interpretation.

Results:

Eighteen of the 24 Q sorts significantly and uniquely defined three factors, which together explained 38% of the study variance. Factor 1, “Cautiously Concerned” (defined by 10 Q sorts; eigenvalue 5.51; 23% variance), recognized AI’s potential benefits but emphasized accuracy, safety, professional supervision, and preservation of the physician’s role. Factor 2, “Optimistically Open” (defined by 4 Q sorts; eigenvalue 1.94; 8% variance), emphasized transparency, informational benefits, self-assessment, and preparation for medical consultation. Factor 3, “Privacy Anxious” (defined by 4 Q sorts; eigenvalue 1.67; 7% variance), prioritized protection of personal health information and required stronger data safeguards and professional guidance. Across the three viewpoints, participants were generally cautious about independent AI use, valued alignment with physician advice, and considered data protection important.

Conclusions:

Patients’ perceptions of AI-mediated health information seeking were neither uniformly positive nor negative but reflected different configurations of perceived benefit, trust, privacy, and professional involvement. Patient-facing AI should therefore be designed and implemented with clinical credibility, transparent data practices, appropriate professional oversight, and clear positioning as a complement rather than a substitute for professional care.