Limitations and practical role of large language models in preoperative patient education: An expert-based assessment

INTRODUCTION: To evaluate the quality and educational value of information generated by large language models (LLMs) for use in preoperative counselling with patients undergoing sleeve gastrectomy.
METHODS: Two LLMs were used to generate responses to common patient questions. Bariatric surgeons then evaluated the responses using a Likert scale to assess their accuracy, clarity and usefulness.
RESULTS: Both models mostly produced accurate and clear information on procedure details, though individual risk communication was sometimes limited.
DISCUSSION AND CONCLUSION: LLMs can provide supplementary information for preoperative education, but should not replace discussions with physicians. During preoperative counselling, clinicians should guide patients in interpreting AI-generated content.

Keywords: Artificial intelligence, patient education, large language models, bariatric surgery, healthcare communication