Abstract A028: Automated Rule Synthesis from Literature for Agent-Based Modeling of the Tumor Microenvironment
Bibliographic record
Abstract
Abstract Agent-based models (ABMs) are powerful tools for simulating multicellular dynamics in complex biological systems such as the tumor microenvironment (TME), enabling mechanistic insights into cancer progression, immune responses, and therapeutic outcomes. By representing individual cells as autonomous agents with defined behavioral rules, ABMs allow researchers to explore how local interactions give rise to emergent tissue-level features. The PhysiCell modeling platform has broadened access to ABMs by supporting user-friendly, rule-based configuration of cellular behaviors and microenvironmental conditions using a natural language grammar. The process of defining biologically accurate and comprehensive rules, however, remains a significant bottleneck. This process often requires labor-intensive manual curation of the literature and substantial domain expertise to translate complex and sometimes ambiguous biological findings into formal, simulation-ready logic. This challenge limits the scalability and accessibility of ABMs, particularly for rapidly prototyping models in emerging biological contexts or under tight development timelines. To streamline this process, we developed a large language model (LLM)-powered framework for automated rule synthesis from biomedical literature. Leveraging a custom retrieval-augmented generation (RAG) pipeline built on open-source infrastructure, the system identifies relevant scientific content and converts it into interpretable, simulation-compatible rules. While designed for compatibility with PhysiCell, the framework is platform-agnostic and includes a freely accessible web interface to support broad adoption by researchers across disciplines and experience levels. We benchmarked state-of-the-art LLMs for their ability to generate biologically valid and internally consistent rules from multiple—and occasionally conflicting—sources. We also assessed how the volume and quality of input evidence influence output consistency. The resulting rules closely mirrored expert-defined logic, successfully recovering 43 of 49 rules from a published agent-based model of triple-negative breast cancer. By accelerating the translation of biomedical literature into executable models, this AI-driven approach lowers the barrier to in silico experimentation and promotes broader adoption of agent-based modeling in cancer systems biology and precision oncology. Citation Format: Zachary Sims, Eric Cramer, Daniel Derrick, Laura Heiser, Paul Macklin, Young Hwan Chang. Automated Rule Synthesis from Literature for Agent-Based Modeling of the Tumor Microenvironment [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Artificial Intelligence and Machine Learning; 2025 Jul 10-12; Montreal, QC, Canada. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(13_Suppl):Abstract nr A028.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.014 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.004 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.003 | 0.001 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.013 | 0.004 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".