PATHFinder Agent Tailors Prenatal Care with Conversational AI

PATHFinder Agent (Planner for Appropriate Tailored Healthcare) is an innovative, end-to-end conversational agentic system designed to provide tailored prenatal care. This system aligns with the American College of Obstetricians and Gynecologists (ACOG) guidelines for personalized prenatal care. It functions through a four-stage workflow: patient intake, dynamic interaction to gather health and social context, synthesis of an individualized care plan, and clinician oversight. The agent leverages frontier large language models (LLMs) to facilitate structured dialogue and curate care plans. It also integrates with community resources, such as Michigan 211, to provide comprehensive support. Evaluations using expert-curated rubrics across five clinical dimensions showed that GPT-5.2 achieved the highest average score (77.6%), though the system identified specific gaps, particularly in antenatal testing recommendations. Future validation through human participant studies and randomized controlled trials is planned to further assess its effectiveness and safety.