Elizabeth Bondi-Kelly and Alex Peahl receive Google grant to develop AI-powered reproductive health chatbot

Elizabeth Bondi-Kelly, assistant professor of computer science and engineering at the University of Michigan, and Alex Peahl, assistant professor of obstetrics and gynecology in Michigan Medicine, have received a $75,000 Google Academic Research Award for their project titled “MI-OB-Now: Advancing Access to Reproductive Health Information with Gen-AI Clinical Messaging Systems.” They aim to significantly enhance the availability and quality of reproductive health information, especially for historically marginalized populations.

The Google Academic Research Awards program supports pioneering research in computer science and related fields, funding a select few proposals each year that hold strong potential for significant technological and societal impact.

Bondi-Kelly and Peahl’s project addresses the critical challenge of misinformation and inequity in the domain of reproductive health. By developing MI-OB-Now, a chatbot to answer patients’ pregnancy-related questions, the team seeks to provide patients with reliable, timely information to better support their health needs in this crucial area. This system will use the latest advancements in generative AI and multi-agent models to answer patient inquiries, refer complex or safety-critical questions to clinicians, and counteract misinformation from social media and other digital platforms.

“The current landscape of reproductive health information is rife with inaccuracies and inequities,” explained Bondi-Kelly. “Our goal is to ensure that patients have access to trusted information, while also alleviating some of the burden on clinicians.”

The project will progress through multiple phases: gathering questions from patients, identifying online misinformation, developing the MI-OB-Now system, and conducting a pilot phase at Michigan Medicine. By using both patients’ real-world questions and verified guidelines, the system will provide responses to common, straightforward questions and referrals for more complex queries.

MI-OB-Now will be able to address a variety of straightforward questions that patients often face, from pregnancy concerns like “Can I go swimming while pregnant?” to postpartum inquiries such as “Is ibuprofen safe while breastfeeding?” The system will adapt to different types of queries, leveraging multimodal inputs like text and video to provide accurate and contextually appropriate information.

“Our collaboration ensures that we are not only identifying patients’ most pressing questions, but also developing effective, efficient solutions for answering them,” said Peahl. “In this way, we hope to improve care for birthing people and the health care workers caring for them, particularly in areas with limited access to care.”

Additionally, the project has significant broader impacts. By reducing the frequency of urgent prenatal visits and emergency room presentations, MI-OB-Now can help lessen the in-basket burden on physicians and reduce burnout. The project also promises to create a feedback loop that helps to continuously improve responses and address emerging misinformation trends.

By developing a system that leverages the strengths of AI while ensuring safety and reliability through deferral to clinicians, Bondi-Kelly and Peahl are setting the stage for more effective healthcare communication tools in the future.

“With the support of Google, we are positioned to transform how reproductive health information is accessed and perceived, particularly for those in underserved communities,” noted Bondi-Kelly. “Ultimately, our work aims to empower patients with accurate information and to support clinicians in providing high-quality care.”