Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note3 min read

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

Source

arXiv

arxiv.org/abs/2607.13608v1

Read the source

Source

Plain-Language Brief

The MEDA system is an automated framework using Large Language Models (LLMs) and Symbolic Regression (SR) to discover Ordinary Differential Equations (ODEs) for biological systems. It retrieves background knowledge, defines variables, generates constraints, proposes candidate ODEs, and fits them. The study evaluates the system's performance across various tasks and shows that it can recover correct state variables, achieve strong structural recovery, and produce biologically plausible models.

What the Document Says

  • The MEDA system is introduced for discovering ODE models of biological systems.
  • MEDA uses LLMs and SR.
  • It retrieves background knowledge.
  • DEFINES admissible variables (possibly referring to defining which variables are allowed or relevant).
  • Generates mechanistic constraints.
  • Proposes candidate ODEs.
  • Fits and evaluates them.
  • Evaluation across canonical model retrieval, reasoning-based extrapolation, open-ended discovery (with and without experimental data).
  • Recovered correct state variables.
  • Achieved strong structural recovery in retrieval and extrapolation tasks.
  • Produced biologically plausible discovery-oriented models.
  • Ablation and robustness analyses show that knowledge-guided formalization and mechanistic constraints are essential components.
  • Numerical fitting alone can preserve trajectory-compatible but biologically incorrect equations.

Key Concepts Mentioned

  • MEDA system
  • Large Language Models (LLMs)
  • Symbolic Regression (SR)
  • Ordinary Differential Equations (ODEs)
  • biological systems
  • background knowledge
  • mechanistic constraints
  • candidate ODEs
  • evaluation
  • canonical model retrieval
  • reasoning-based extrapolation
  • open-ended discovery
  • ablation analysis
  • robustness analysis

Claims Made by the Document

  • MEDA can discover ODE models for biological systems.
  • Knowledge-guided formalization and mechanistic constraints are essential components of MEDA.
  • Numerical fitting alone may preserve trajectory-compatible but biologically incorrect equations.

Evidence Used

  • The introduction and evaluation of the MEDA system in the text.

Methods / Approach

  • Use of LLMs and SR
  • Retrieval of background knowledge
  • Definition of admissible variables
  • Generation of mechanistic constraints
  • Proposal of candidate ODEs
  • Fitting and evaluation of ODEs

Limitations Stated by the Authors

Not stated in source

Limitations Noted by the Agent

Agent interpretation — not from the source.

  • The study only evaluates the performance of MEDA on specific tasks. It is unknown how it performs on other tasks or real-world applications.
  • The text does not provide details about the complexity or scalability of the MEDA system.

Open Questions

  • What are the limitations and potential improvements for the MEDA system?
  • How well does the MEDA system perform in real-world applications beyond the tasks evaluated in this study?

Notes for Future Q&A

May include interpretation. Labeled where it does.

  • The focus of this research is on demonstrating the capabilities of the MEDA system, so there may be limitations or challenges not addressed within the study.

Raw Abstract / Excerpt

Automatic scientific discovery has long been a goal of computational scholars - a machine that can discover nature's secrets on its own, moving computational systems beyond data-fitting tools toward the generation and refinement of mechanistic models of the universe. Recent advances in symbolic regression (SR) and large-language-model (LLM)-based agents suggest that such systems can recover equations from data, incorporate domain priors, and automate parts of the research workflow. However, most existing approaches either focus on narrow equation-discovery benchmarks or broad end-to-end automation pipelines, while biological systems remain comparatively underexplored. Here, we introduce the MEDA system, an LLM- and SR-powered agentic framework for discovering ordinary-differential-equation (ODE) models of biological and biologically inspired dynamical systems. MEDA retrieves background knowledge, defines admissible variables, generates mechanistic constraints, proposes candidate ODEs, and fits and evaluates them. We evaluate it across canonical model retrieval, reasoning-based extrapolation to unseen variants, and open-ended discovery, with and without experimental data. Across these settings, MEDA recovered the correct state variables, achieved strong structural recovery in retrieval and extrapolation tasks, and produced biologically plausible discovery-oriented models. Ablation and robustness analyses show that knowledge-guided formalization and mechanistic constraints are load-bearing components, whereas numerical fitting alone can preserve trajectory-compatible but biologically incorrect equations.