EpidemIQs: Prompt-to-Paper LLM Agents for Epidemic Modeling and Analysis
A new multi-agent LLM framework autonomously conducts full epidemic studies, generating scientific papers for under $2.
A team of researchers has published a paper on arXiv introducing EpidemIQs, a novel multi-agent LLM framework designed to automate the complex, interdisciplinary process of epidemic modeling and analysis. The system represents a significant step toward AI-driven scientific research, taking a user's initial prompt and autonomously guiding it through five predefined research phases—from literature review and analytical derivation to network modeling, stochastic simulations, and final documentation. The goal is to accelerate research in fields like epidemiology and network science by handling tasks that traditionally require extensive human expertise and time.
The framework employs two types of agents: a 'scientist' agent for high-level planning and coordination, and specialized 'task-expert' agents that execute specific duties like running simulations or creating visualizations. Powered by GPT-4.1 and the more efficient GPT-4.1 Mini, the system completed full studies with an average total token usage of 870,000 at a cost of approximately $1.57 per report. In evaluations, EpidemIQs addressed various epidemic scenarios with a 79% average task success rate, outperforming a comparable iterative single-agent approach. This demonstrates a practical, cost-effective path for automating technical research workflows and generating structured scientific manuscripts.
- Multi-agent framework autonomously runs the full research workflow from literature review to final manuscript.
- Costs about $1.57 per study using GPT-4.1 models, averaging 870K tokens per complete report.
- Achieved a 79% task success rate in evaluations, outperforming a single-agent iterative baseline.
Why It Matters
Dramatically lowers the cost and time barrier for complex epidemic modeling, enabling faster, data-driven public health responses.