DiagnoLLM: A Hybrid Bayesian Neural Language Framework for Interpretable Disease Diagnosis

TL;DR

DiagnoLLM is a hybrid AI framework combining Bayesian models, eQTL-guided deep learning, and LLM-based narrative generation to provide interpretable and trustworthy disease diagnosis, demonstrated in Alzheimer's detection.

Contribution

The paper introduces DiagnoLLM, a novel hybrid framework that integrates biological modeling, deep learning, and language models for interpretable clinical diagnosis.

Findings

Achieved 88.0% accuracy in Alzheimer's detection.

LLMs effectively generate accurate, actionable diagnostic reports.

Hybrid approach enhances interpretability and trust in clinical AI.

Abstract

Building trustworthy clinical AI systems requires not only accurate predictions but also transparent, biologically grounded explanations. We present \texttt{DiagnoLLM}, a hybrid framework that integrates Bayesian deconvolution, eQTL-guided deep learning, and LLM-based narrative generation for interpretable disease diagnosis. DiagnoLLM begins with GP-unmix, a Gaussian Process-based hierarchical model that infers cell-type-specific gene expression profiles from bulk and single-cell RNA-seq data while modeling biological uncertainty. These features, combined with regulatory priors from eQTL analysis, power a neural classifier that achieves high predictive performance in Alzheimer's Disease (AD) detection (88.0\% accuracy). To support human understanding and trust, we introduce an LLM-based reasoning module that translates model outputs into audience-specific diagnostic reports, grounded in clinical features, attribution signals, and domain knowledge. Human evaluations confirm that these reports are accurate, actionable, and appropriately tailored for both physicians and patients. Our findings show that LLMs, when deployed as post-hoc reasoners rather than end-to-end predictors, can serve as effective communicators within hybrid diagnostic pipelines.