An Agentic Generative Large Language Model for Treatment Planning of Colorectal Cancer

2026-08-10

AI Summary

This study addresses the challenges of heterogeneous data integration, dynamic adherence to clinical guidelines, and safe decision-making in precision treatment for colorectal cancer by proposing GatorOnco—the first large language model that integrates agent-based reasoning with large-scale domain adaptation. The approach leverages domain-adaptive pretraining, model fusion, two-stage post-training, agent-based reinforcement learning, and retrieval-augmented generation (RAG) to enable dynamic incorporation of clinical guidelines and generate safe, controllable treatment plans. In blinded evaluations, GatorOnco significantly outperformed existing open-source large language models (P<0.01), surpassing human experts in readability and completeness while matching oncologists in correctness, timeliness, and safety.

Abstract

Treatment planning in precision oncology requires synthesizing heterogeneous patient information with rapidly evolving clinical guidelines to ensure guideline-concordant care. While large language models (LLMs) show promise in many diagnostic tasks, their adoption for high-stakes treatment planning is hindered by complex reasoning, adherence to timely clinical guidelines, and safety concerns. In this study, we present GatorOnco, an agentic LLM for colorectal cancer (CRC) treatment planning. GatorOnco is developed using a total of 282 billion tokens of biomedical text, including healthcare system-scale clinical text comprising 166 billion tokens from UF Health. We implemented a domain-adaptation method that integrates pre-training, model merging, a two-stage post-training approach, and agent-based reinforcement learning. An agentic retrieval-augmented generation (RAG) approach dynamically integrates time-sensitive clinical guidelines into the reasoning process. In a blind, randomized clinical evaluation conducted by five UF Health oncologists, GatorOnco significantly outperformed open-source LLMs (P < 0.01) and achieved expert-level performance comparable to UF Health oncologists. Compared with expert oncologists, GatorOnco received significantly higher ratings for readability (4.46 vs. 4.19, P < 0.01) and completeness (3.91 vs. 3.52, P < 0.01), while showing statistically comparable performance in correctness (4.09 vs. 4.11, P = 0.921), currency (4.04 vs. 3.98, P = 0.478), and safety (4.22 vs. 4.22, P = 0.999). These findings demonstrate that integrating agentic reasoning with large-scale domain adaptation can help bridge the gap for generative AI in high-stakes cancer treatment planning.

Problem

• treatment planning
• colorectal cancer
• clinical guidelines
• large language models
• precision oncology

Innovation

• Agentic LLM
• Retrieval-Augmented Generation (RAG)
• Domain Adaptation
• Treatment Planning
• Colorectal Cancer