A Large Language Model-based Approach for Analyzing Covariates of Health Equity in Registered Research Projects

Abstract Large language models (LLMs) have made significant advancements in natural language processing, offering broad applications in multiple domains. This study explores the use of the GPT-3.5 LLM to conduct efficient and robust computational analysis of registered research projects on the All of Us platform. Specifically, we explore the association between projects pursuing health equity research and: the project’s use of demographic categories (which All of Us enables), the multi-institutional composition of the team leading the project, and the involvement of R2 institutions (compared to only R1 institutions). We demonstrate the utility of GPT-3.5 in automating tasks ranging from generating Python scripts for extracting attributes from free text (such as project description and goals) to identifying and classifying institutions as R1 and R2, and summarizing project details into Unified Medical Language System (UMLS)-coded medical keywords. These contributions significantly reduced manual workload, allowing researchers to focus on more in-depth analysis. Our results reveal health equity insights not readily available in the original All of Us research hub. Specifically, we find a strong positive association between the use of demographic data and projects focused on health equity, while other associations such as health equity projects conducted by institutions were positive but weaker and more dependent on specific project topics. Competing Interest Statement The authors have declared no competing interest. Funding Statement The author(s) received no specific funding for this work. Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: This study does not meet the criteria for human subjects research. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes Data Availability All data is publicly available, and linked in the manuscript. https://github.com/navapatn/all-of-us-projects/blob/main/data/augmented-data/processed_responses.csv Abbreviations - AI - Artificial Intelligence - CI - Confidence Interval - CSV - Comma Separated Values - GPT - Generative Pre-trained Transformers - HTML - HyperText Markup Language - LLM - Large Language Model - OR - Odds Ratio - UMLS - Unified Medical Language System - U.S. - United States