KG-TRACE: A Neuro-Symbolic Framework for Mechanistic Grounding in Antimicrobial Resistance Prediction

2026-06-24

AI Summary

Current genome-wide sequencing–based antimicrobial resistance prediction models suffer from limited clinical credibility due to insufficient alignment with established biological mechanisms. This work proposes KG-TRACE, a novel framework that integrates the WHO mutation knowledge graph with neural genomic models for the first time, enabling mechanistically interpretable predictions through a learnable cognitive trust gate that dynamically fuses data-driven evidence with symbolic biological knowledge. The study introduces the Biological Grounding Ratio (BGR) to quantify consistency between neural attributions and prior knowledge, while enhancing clinical auditability via symbolic coverage and uncertainty tagging. Evaluated on the CRyPTIC *Mycobacterium tuberculosis* cohort, the model achieves an AUROC of 0.9760 for isoniazid resistance prediction with 92.5% symbolic coverage and effectively identifies spurious co-occurrence patterns in multidrug resistance, offering verifiable support for clinical decision-making.

Abstract

While WGS-based AMR prediction has reached high accuracy, existing models lack a mechanism to ground neural attributions in established biological pathways. We present KG-TRACE, a novel neuro-symbolic framework that integrates the WHO mutation knowledge graph (KG) as a structured biological constraint on a neural genomic model. Unlike existing methods that learn statistical patterns in isolation, KG-TRACE fuses genomic features and RotatE-based KG embeddings through a learned epistemic trust gate, dynamically weighting neural evidence against symbolic biological knowledge. Evaluated on the CRyPTIC M. tuberculosis cohort, KG-TRACE achieves an AUROC of 0.9760 for isoniazid, achieving competitive accuracy while its primary value lies in symbolic grounding, not predictive uplift. More importantly, we introduce the Biological Grounding Ratio (BGR), a dataset-level metric that quantifies alignment between neural attributions and established biology. Our framework achieves a 92.5% symbolic coverage of isoniazid-resistant predictions and effectively identifies MDR co-occurrence artifacts by issuing laboratory follow-up flags for 'UNCERTAIN' cases. We demonstrate that neuro-symbolic grounding provides a verifiable audit trail for clinicians, bridging the gap between predictive accuracy and clinical trust.

Problem

• antimicrobial resistance
• neural attribution
• biological grounding
• knowledge graph
• clinical interpretability

Innovation

• neuro-symbolic
• knowledge graph
• biological grounding
• epistemic trust gate
• antimicrobial resistance