Reproduce “SynthAgent: A Multi-Agent LLM Framework for Realistic Patient Simulation -- A Case Study in Obesity with Mental Health Comorbidities” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Arman Aghaee, Sepehr Asgarian, Jouhyun Jeon

SynthAgent is a multi-agent LLM framework that generates realistic synthetic patient profiles, initially focusing on obesity with mental health comorbidities, by integrating diverse empirical data and personalized personality traits. The framework achieved mean quality scores above 76 using GPT-5 and Claude 4.5 Sonnet as core LLM engines, with Claude 4.5 Sonnet also demonstrating superior diversity in patient narratives.

Key insights

  • Large Language Models (LLMs) can effectively serve as core reasoning engines within a multi-agent framework to generate complex, coherent, and realistic synthetic patient profiles.
  • Integrating diverse empirical data sources and comprehensive psychological dimensions into patient simulations significantly enhances their realism and clinical coherence.
  • A multi-agent architecture supports structured refinement and automated evaluation, ensuring logical consistency, medical plausibility, and overall quality in generated synthetic data.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.