Improving CKD Clinical Trial Representation with Educational Videos and Generative Artificial Intelligence

Abstract

Background: The prevalence of Chronic Kidney Disease (CKD) among underserved communities means representative study cohorts are a requirement for scientifically robust and equitable CKD clinical trials. However, despite the importance of representative cohorts and increased attention on the problem, CKD trials remain largely unrepresentative. This is largely due to the many barriers to trial enrollment faced largely by Hispanic and African-American CKD communities – things like logistics (eg, dialysis centers being more accessible than trial sites to many urban minorities), distrust of the healthcare system, and research unawareness. Efforts like peer-based recruitment & research literacy education have been promising in an effort to address those barriers and increase representative trials. Methods: Here, we present a proof-of-concept showing how Generative AI (GenAI) can be used to address barriers to enrollment faced by diverse communities. Specifically, we show how a set of GenAI models produces 120-second videos with CKD clinical trials information, where the videos are tailored to be accessible, inclusive, and ultimately more meaningful to different local demographic CKD patients. In this research, we show a case study in which our GenAI algorithms were trained and tuned to specifically create educational videos for a ESRD study whose protocol and Informed Consent Form were downloaded from clinicaltrials.gov. Model training and tuning was based on open source GenAI models, academic research literature, public & nonpublic structured data, and subject matter expertise from Clinical Research Coordinators, IRB reviewers, and sensitivity readers. Results: As a case study, we generated three semi-personalized videos educating about a study of Empagliflozin in ESRD patients; in each video, the GenAI adapts tone, language, visual style, and content emphasis for a unique ESRD patient profile – a 65 year old African-American female in rural Alabama, a 35 year old African-American male in Chicago, and a 25 year old Hispanic male in suburban Tampa, Florida. Conclusion: We highlight how the GenAI’s recommendations – like defining what clinical research is and why it is important for the elderly patient profile vs. describing onsite childcare options for the younger Hispanic profile – helps address underserved communities’ barriers to trial enrollment more effectively. Funding: Commercial Support - Tea Leaf Health, Inc

Ähnliche Arbeiten