Commentary: Can large language models improve the quality and efficiency of health policy surveillance?

Abstract

Legal epidemiology is a growing field wherein researchers examine the effects of health policies on health outcomes. Policy surveillance is a key part of the legal epidemiology process and leads to the creation of a legal dataset for econometric analyses. Policy surveillance includes conceptualizing the research scope, collecting legal data, and coding data to answer legal questions. Historically, policy surveillance has been slow and time-consuming, limiting timely information about the effects of health policies on health outcomes. Researchers are exploring the role of artificial intelligence using large language models (LLMs) to increase the efficiency and quality of the policy surveillance process. In this commentary, we describe potential uses of LLMs by researchers conducting policy surveillance as well as potential challenges that will need to be addressed before LLMs are widely used in the process. We also suggest areas for future research, while emphasizing the need for development of guidelines for LLM use in policy surveillance.