Automatic Ordinary Differential Equations Discovery For Biological Systems Using Large Language Model Powered Agentic System

Computational scholars aim for automatic scientific discovery to move beyond data-fitting toward generating mechanistic models of the universe. Recent progress in symbolic regression (SR) and large-language-model (LLM)-based agents enables systems to recover equations from data, integrate domain priors, and automate research workflows. While many current methods target narrow benchmarks or broad pipelines, biological systems remain underexplored. Consequently, a new LLM-powered agentic system is introduced to automate Ordinary Differential Equations discovery specifically for biological systems.