Answers are grounded in this paper and cite their sources.
Background: ChatGPT is a popular open-source large language model (LLM) that uses supervised learning to create human-like queries. In recent years, ChatGPT has generated excitement in the medical field. However, its accuracy must be carefully evaluated to determine its usefulness in patient care. In this literature review, the authors examine whether ChatGPT can accurately answer frequently asked questions (FAQs) from patients, make clinical recommendations, and effectively categorize patient symptoms. Methods: A database search in PubMed was conducted using the search terms "ChatGPT," "accuracy," and "clinical decision-making," yielding 122 unique references. Two screening stages resulted in 9 studies that met the evaluation criteria for this review. Results: Analysis of 9 studies showed that while ChatGPT can answer FAQs, offer recommendations, and categorize symptoms in less complicated scenarios, its clinical accuracy ranged from 20% to 95%. ChatGPT may be helpful in specific clinical scenarios; however, its variable accuracy makes it unsuitable as a stand-alone point-of-care product. Conclusions: ChatGPT is only adept at providing generalized recommendations when individual patient care is more suitable. Further research is needed to identify where ChatGPT delivers the most accurate responses and how it can supplement traditional care.
No takes yet. Share an insight, caveat, or question.
Geracitano et al. (2025) studied this question.