Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

NeuroSymAD: A Neuro-Symbolic Framework for Interpretable Alzheimer's Disease Diagnosis

Yexiao He, Ziyao Wang, Yuning Zhang, Tingting Dan, Tianlong Chen, Guorong Wu, Ang Li

TL;DR

Neural diagnosis of Alzheimer's disease benefits from multi-modal data but existing deep models are often black-box and limited in incorporating clinical knowledge. NeuroSymAD addresses this by uniting a neural MRI perception module with a symbolic reasoning module that uses rules $\mathcal{R}$ generated via a large language model with Retrieval-Augmented Generation (LLM-RAG), all trained end-to-end. A dedicated automated knowledge acquisition module continuously builds and updates $\mathcal{R}$ from guidelines and literature, enabling interpretable, rule-guided adjustments to the neural predictions expressed as logits $\mathbf{y}_i$. On the ADNI dataset, NeuroSymAD with a 3D ResNet backbone achieves higher accuracy and F1-score than state-of-the-art baselines while providing explanatory reports and region-aware heatmaps, illustrating both superior performance and transparent reasoning for clinical deployment.

Abstract

Alzheimer's disease (AD) diagnosis is complex, requiring the integration of imaging and clinical data for accurate assessment. While deep learning has shown promise in brain MRI analysis, it often functions as a black box, limiting interpretability and lacking mechanisms to effectively integrate critical clinical data such as biomarkers, medical history, and demographic information. To bridge this gap, we propose NeuroSymAD, a neuro-symbolic framework that synergizes neural networks with symbolic reasoning. A neural network percepts brain MRI scans, while a large language model (LLM) distills medical rules to guide a symbolic system in reasoning over biomarkers and medical history. This structured integration enhances both diagnostic accuracy and explainability. Experiments on the ADNI dataset demonstrate that NeuroSymAD outperforms state-of-the-art methods by up to 2.91% in accuracy and 3.43% in F1-score while providing transparent and interpretable diagnosis.

NeuroSymAD: A Neuro-Symbolic Framework for Interpretable Alzheimer's Disease Diagnosis

TL;DR

Neural diagnosis of Alzheimer's disease benefits from multi-modal data but existing deep models are often black-box and limited in incorporating clinical knowledge. NeuroSymAD addresses this by uniting a neural MRI perception module with a symbolic reasoning module that uses rules generated via a large language model with Retrieval-Augmented Generation (LLM-RAG), all trained end-to-end. A dedicated automated knowledge acquisition module continuously builds and updates from guidelines and literature, enabling interpretable, rule-guided adjustments to the neural predictions expressed as logits . On the ADNI dataset, NeuroSymAD with a 3D ResNet backbone achieves higher accuracy and F1-score than state-of-the-art baselines while providing explanatory reports and region-aware heatmaps, illustrating both superior performance and transparent reasoning for clinical deployment.

Abstract

Alzheimer's disease (AD) diagnosis is complex, requiring the integration of imaging and clinical data for accurate assessment. While deep learning has shown promise in brain MRI analysis, it often functions as a black box, limiting interpretability and lacking mechanisms to effectively integrate critical clinical data such as biomarkers, medical history, and demographic information. To bridge this gap, we propose NeuroSymAD, a neuro-symbolic framework that synergizes neural networks with symbolic reasoning. A neural network percepts brain MRI scans, while a large language model (LLM) distills medical rules to guide a symbolic system in reasoning over biomarkers and medical history. This structured integration enhances both diagnostic accuracy and explainability. Experiments on the ADNI dataset demonstrate that NeuroSymAD outperforms state-of-the-art methods by up to 2.91% in accuracy and 3.43% in F1-score while providing transparent and interpretable diagnosis.

Paper Structure

This paper contains 8 sections, 4 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Literature
  3. Methods
  4. Neural Perception Module
  5. Symbolic Reasoning Module
  6. Automated Knowledge Acquisition Module
  7. Experiments
  8. Conclusion

Figures (3)

  • Figure 1: Overview of NeuroSymAD
  • Figure 2: An example of NeuroSymAD's diagnostic process and the generated explanatory report, demonstrating how the symbolic reasoning module corrects the misclassification of neural network.
  • Figure 3: Heatmaps (used to guide the diagnosis for labeling AD) overlaid on MRIs for two typical AD subjects, showing the activation distribution of a NeuroSymAD model.