Supporting Reproductive Health Equity: Understanding Access to Reliable Health Information in the Age of Generative AI

We face a maternity care crisis in the US: birthing people today are more likely to die in pregnancy than the generation before, with people who are Black or living on low incomes 2-5 times more likely to suffer severe maternal morbidity and mortality. Leading maternity care organizations including the CDC and ACOG have called for increased access to reproductive health information to improve these unacceptable outcomes. Yet, growing maternity care deserts—areas with no maternity care—and the clinician burnout crisis have eroded access to high quality information and care. We aim to build MI-OB-Now, a multi-agent large language model (LLM) system, to answer patient questions automatically within an electronic health record system. We will research multiple safety measures, such as referring key questions to a clinician and addressing potential misinformation, work with communities to ensure that the proposed features meet their needs, and pilot this system with Michigan Medicine (MM). MI-OB-Now will improve access to reproductive health information, particularly for historically marginalized populations, while advancing approaches to providing health information to patients outside a clinical setting, as well as LLM deployment research. 

In this session, we are hoping to receive feedback from the e-HAIL community on work completed so far, proposed next steps, and potential funding avenues.

Zoom information will be sent to e-HAIL members.