Contextual Invertible World Models: A Neuro-Symbolic Agentic Framework for Colorectal Cancer Drug Response

2026-03-01

AI Summary

This study addresses the challenges in precision oncology posed by high-dimensional genomic data with limited drug response samples (small N, large P) and the lack of interpretability and causal mechanisms in existing deep learning models. The authors propose a neuro-symbolic agent framework that integrates an invertible world model with a large language model–based reasoning layer, incorporating clinical context such as microsatellite instability (MSI) status to enable interpretable prediction of drug response in colorectal cancer. A novel “backward reasoning” mechanism is introduced to simulate CRISPR perturbations, distinguishing actionable therapeutic opportunities from context-dependent resistance. Evaluated on the GDSC dataset (N=83), the method achieves a prediction correlation of r=0.504 (p=0.023), significantly outperforming baseline approaches, with validation from both biological and clinical evidence, thereby advancing the application of interpretable AI in cancer research.

Technology Category

• Machine Learning: Neuro-Symbolic Learning
• Multiagent Systems: Adversarial Agents
• Cognitive Modeling & Cognitive Systems: Agent Architectures

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Precision oncology is currently limited by the small-N, large-P paradox, where high-dimensional genomic data is abundant, but high-quality drug response samples are often sparse. While deep learning models achieve high predictive accuracy, they remain black boxes that fail to provide the causal mechanisms required for clinical decision-making. We present a Neuro-Symbolic Agentic Framework that bridges this gap by integrating a quantitative machine learning World Model with an LLM-based agentic reasoning layer. Our system utilises a forensic data pipeline built on the Sanger GDSC dataset (N=83), achieving a robust predictive correlation (r=0.504) and a significant performance gain through the explicit modelling of clinical context, specifically Microsatellite Instability (MSI) status. We introduce the concept of Inverse Reasoning, where the agentic layer performs in silico CRISPR perturbations to predict how specific genomic edits, such as APC or TP53 repair, alter drug sensitivity. By distinguishing between therapeutic opportunity and contextual resistance, and validating these findings against human clinical data (p=0.023), our framework provides a transparent, biologically grounded path towards explainable AI in cancer research.

Problem

• precision oncology
• small-N large-P
• drug response prediction
• explainable AI
• causal mechanisms

Innovation

• Neuro-Symbolic AI
• Invertible World Models
• Inverse Reasoning
• Microsatellite Instability
• CRISPR Perturbation