Usability and feasibility of a Socratic LLM-supported learning tool for clinical reasoning in undergraduate nursing education.

AIM: To evaluate clinical reasoning outcomes and the usability and feasibility of an AI-supported platform that uses a large language model (LLM) to foster higher-order thinking via the Socratic method. DESIGN: Prospective observational process evaluation guided by the UK Medical Research Council (MRC) and National Institute for Health and Care Research (NIHR) frameworks, conducted alongside outcome evaluation. METHODS: Full-time Bachelor of Science nursing students who used the AI-supported platform between February and May 2025 were recruited. The intervention integrated the ADPIE nursing process (Assessment, Diagnosis, Planning, Implementation, Evaluation) and the clinical reasoning cycle via adaptive Socratic questioning. RESULTS: A total of 142 students completed the evaluation. Most used the platform as a supplementary learning tool, particularly to support case management, review clinical procedures, and answer queries during clinical sessions. Students reported improvements in perceived clinical reasoning, especially in analyzing errors, adopting multiple perspectives, and reflecting on plans prior to intervention. Most participants (74.0%) agreed the platform helped them use ADPIE to guide thinking, and many perceived support for applying clinical reasoning frameworks to decision-making. CONCLUSION: A customized AI-supported platform using an LLM through the Socratic method is a feasible, usable adjunct to traditional nursing education and shows promise for enhancing clinical reasoning. Thoughtful design, grounded in nursing frameworks and dialogic guidance may help encourage deeper learning and discourage a surface-level approach associated with conventional question-answer LLM use.