JMIR Publications

ABSTRACT

Background:

Previous research on patient experiences with generative artificial intelligence (GenAI) has primarily relied on static technology acceptance models to evaluate the use of AI tools in healthcare. However, it has been poorly studied which mechanisms shape acceptance, empowerment, resistance, and the changing relationship with healthcare professionals.

Objective:

This Systematic Literature Review (SLR) evaluates patient experience with GenAI in healthcare, thereby aiming to understand the patient– GenAI interaction in healthcare and frame the dynamics in patient experience.

Methods:

The review consists of empirical studies published in 2022–2026 across six major databases and records. It is conducted following the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA-2020) guidelines. It synthesised diverse methodologies and tools. Records were screened using Rayyan. Researchers assessed study quality using the Mixed Methods Appraisal Tool (MMAT), evaluated theme confidence using the Confidence in the Evidence from Reviews of Qualitative Research (GRADE-CERQUAL) framework, coded data with MAXQDA, and synthesised data using the Thomas and Harden (2008) framework.

Results:

Key findings suggest differing patterns of Patient – GenAI interactions according to their control over GenAI, the authority the GenAI assumes, experience, agency and literacy of the patient. Across studies, prior experience and health/AI literacy were associated with differences in perceptions of empathy, privacy, trust, and clinical autonomy, although the direction and strength differed by context. Furthermore, patients appear to value conversational anonymity while often having limited awareness of data handling. As GenAI tools transition from administrative tasks to clinical operations, some patients become more selective or resistant to the autonomous clinical roles, favouring the human-to-human diagnostic relation. The review proposes the Dual-Pathway Patient-GenAI Conceptual Model (DP-PGCM) to map patient-controlled and provider-deployed/controlled interaction trajectories.

Conclusions:

Patients’ GenAI use is a dynamic process. It is role-dependent rather than technology-wide. Patient experience appears to be structured by who controls the tool/system, what clinical authority the tool is granted, and how patients interpret its empathy, privacy, and epistemic reliability. Clinical Trial: PROSPERO registration number: (CRD420261362444). The title and keywords are revised to include the conceptual model.