JMIR Publications

ABSTRACT

Large language models are exhibiting remarkable performance in clinical contexts, with exemplar results ranging from expert-level attainment in medical examinations to superior accuracy and relevance when responding to patient queries than real doctors on a social media website. Deployment of large language models in conventional healthcare settings is yet to be reported, and there remains an open question as to what evidence should be required before such deployment is warranted. Early validation studies use unvalidated surrogate variables to represent clinical aptitude, and it may be necessary to conduct prospective randomised-control trials to justify use of a large language model for clinical advice or assistance as potential pitfalls and pain-points cannot be exhaustively predicted. As large language models continue to revolutionise the field, there is an opportunity to improve the rigour of artificial intelligence research to reward innovation resulting in real benefit to real patients.