EndoGov: A knowledge-governed multi-agent expert system for endometrial cancer risk stratification

TL;DR

EndoGov is a multi-agent expert system that combines structured evidence extraction with rule-based governance to improve guideline-compliant endometrial cancer risk stratification, demonstrating high accuracy and safety in multiple datasets.

Contribution

The paper introduces EndoGov, a novel two-tier multi-agent system integrating rule-based governance with evidence extraction for improved clinical decision-making in EC risk stratification.

Findings

Achieved 0.943 accuracy and 0.973 macro AUC on TCGA-UCEC dataset.

Supported guideline-derived labels with 0.842 accuracy on CPTAC-UCEC, outperforming neural baselines.

Localized residual failures mainly to upstream molecular detection rather than governance.

Abstract

Multimodal artificial intelligence models for endometrial cancer (EC) risk stratification typically optimize aggregate predictive performance but provide limited mechanisms for enforcing mandatory guideline overrides, such as assigning POLE-mutated tumors to the low-risk group despite high-grade morphology. We present EndoGov, a two-tier multi-agent expert system that factorizes the decision process as D(x) = G(P(x), R), where specialist agents P extract structured evidence and a governance agent G applies an executable rule set R. Tier 1 comprises pathology, molecular, and clinical agents that independently generate schema-constrained reports from frozen foundation-model features or structured records. Tier 2 queries an evidence-level-weighted Guideline Knowledge Graph, using deterministic hard-path rules for high-priority overrides and constrained soft-path reasoning for ambiguous cases. In TCGA-UCEC (n=541), EndoGov achieved 0.943 accuracy, 0.973 macro AUC, and a conditional logic-violation rate (C-LVR) of 0.93% among trigger-exposed cases. In CPTAC-UCEC (n=95), where reference labels are guideline-derived, EndoGov reached 0.842 accuracy compared with < 0.31 for locked-transfer neural baselines, supporting governance-pathway transfer under distribution shift rather than validation against independent clinical truth. End-to-end safety decomposition localized residual failures primarily to upstream molecular detection rather than downstream governance. Backend-swap experiments further showed that hard-path compliance is invariant to the LLM backend. These findings indicate that explicit clinical-rule governance can provide guideline-compliant, auditable EC risk assignment while preserving competitive discrimination.