This research presents a novel approach using large language models to simulate human behavior in healthcare settings with limited historical data.

• Creates realistic simulations of maternal health program participants without extensive datasets
• Models how beneficiaries interact with health information messages
• Includes uncertainty estimation for more reliable decision-making
• Demonstrates practical application for public health interventions

This innovation enables healthcare providers to test and optimize interventions before deployment in real-world settings, potentially improving maternal health outcomes in resource-constrained environments.

LLM-based Agent Simulation for Maternal Health Interventions: Uncertainty Estimation and Decision-focused Evaluation