Combined Use of Topic-Analysis and LLMs on Online Peer-to-Peer Medicaid Discussions to Detect the Impact of HR1 on Individuals' Medicaid Experience Across The US

Medicaid is a government program providing benefits to 1 in 5 people in the US, including medical care for children and adults, long-term care for the elderly, and home/community-based services for the disabled[1]. Effective January 1st 2027, HR1 (One Big Beautiful Bill) passed by Congress will implement new administrative requirements to Medicaid expected to strip benefits from 1.8 million adults, children, and seniors[2]. As each US state administers Medicaid independently, the timing, frequency, and severity of those losses may vary across states. Monitoring that human impact requires an detailed ongoing view of affected individuals and their families. To monitor the human-impact of HR1, this work develops a proof-of-concept using topic analysis and LLMs to generate detailed categorization of individuals′ Medicaid experiences across multiple states and over time.