US Occupational Medicine Clinicians' Perceptions and Practices With Respect to Artificial Intelligence Large Language Models: A Mixed-Methods Investigation.

Abstract

The aim of the study was to explore US occupational and environmental medicine (OEM) clinicians' perceptions, knowledge, practices, and interest surrounding large language models (LLMs). An online survey and semistructured interviews were conducted between April 2024 and July 2025 with a sample of US OEM clinicians. Quantitative and qualitative data analyses were performed. There were 60 survey respondents and 10 interviewees. Most respondents reported that they do not currently use LLMs in their clinical practice (70.0%, n = 42). Composite trust scores significantly predicted intention to use LLMs ( B = 0.57, P = 0.019, 95% CI [0.10, 1.03]). The interview data converged with and complemented the survey findings. Although most OEM clinicians in this sample reported not using LLMs in clinical practice, the majority expressed an interest, with trust being a significant predictor of intention to use LLMs.