Large Language Models in Patient Education and Clinical Communication: Reliability, Safety, and Oversight

Abstract

Background: The adoption of clinical language models in patient education, discharge instructions, and clinical correspondence creates opportunities to improve clinical services, but technical performance alone does not establish patient benefit. Objective: To critically examine applications, evidence, safety, equity, and implementation of clinical language models. Approach: This narrative review develops an analytical synthesis of established clinical AI principles and reporting guidance; it does not report a systematic search, a new dataset, or an empirical trial. Findings: Key opportunities include plain-language explanation, multilingual support, and document summarization. Important limitations include hallucination, outdated evidence, and misleading fluency, as well as unsafe advice, omission of warnings, and disclosure of sensitive data. Responsible adoption requires source attribution, escalation pathways, and mandatory human approval, transparent reporting, and evaluation of factual accuracy, readability, actionability, comprehension, and harm. Conclusion: The clinical value of clinical language models depends on fit with the care pathway, prospective evaluation, accountable human oversight, and sustained monitoring after deployment.