JMIR Publications

ABSTRACT

Background:

Large language models(LLMs)are increasingly used to support patient health education. However, little is known about how frontline nurses, who deliver most patient education, experience these tools in clinical practice or what benefits and risks emerge from their use.

Objective:

To explore frontline nurses’ experiences of using LLM-assisted patient education and examine the practical tensions and implementation challenges associated with human–LLM collaborative health education.

Methods:

Semi-structured interviews were conducted with 16 hematological nurses (from December 2025 to March 2026) regarding their human-machine collaborative practices. Data were analyzed using inductive thematic analysis, with two researchers independently coding the data and reaching consensus through discussion.

Results:

Nurses’ experiences were characterized by three major tensions. First, LLMs improved efficiency and content standardization but raised safety concerns and oversight burden, as nurses remained responsible for guiding, reviewing, and correcting LLM generated content. Second, LLMs use simultaneously supported professional learning while increasing perceived responsibility and competence demands. Third, patients benefited from specialized information but faced barriers including low health literacy and technological unfamiliarity. Across these tensions, nurses’ value lay not merely in using LLMs, but in safely integrating them into clinical workflows while managing risks and maintaining accountability.

Conclusions:

LLMs can improve the efficiency and accessibility of patient health education, benefiting both nurses and patients. However, their use also introduces practical tensions related to safety, professional responsibility, and patient applicability. Future implementation and design of these models should better support nurse oversight, support contextual adaptation, and promote patient-centered care. What is already known? Clinical Trial: Trial Registration: ChiCTR2500107909