Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AD-Reasoning: Multimodal Guideline-Guided Reasoning for Alzheimer's Disease Diagnosis

Qiuhui Chen, Yushan Deng, Xuancheng Yao, Yi Hong

Abstract

Alzheimer's disease (AD) diagnosis requires integrating neuroimaging with heterogeneous clinical evidence and reasoning under established criteria, yet most multimodal models remain opaque and weakly guideline-aligned. We present AD-Reasoning, a multimodal framework that couples structural MRI with six clinical modalities and a rule-based verifier to generate structured, NIA-AA-consistent diagnoses. AD-Reasoning combines modality-specific encoders, bidirectional cross-attention fusion, and reinforcement fine-tuning with verifiable rewards that enforce output format, guideline evidence coverage, and reasoning--decision consistency. We also release AD-MultiSense, a 10,378-visit multimodal QA dataset with guideline-validated rationales built from ADNI/AIBL. On AD-MultiSense, AD-Reasoning achieves state-of-the-art diagnostic accuracy and produces structured rationales that improve transparency over recent baselines, while providing transparent rationales.

AD-Reasoning: Multimodal Guideline-Guided Reasoning for Alzheimer's Disease Diagnosis

Abstract

Alzheimer's disease (AD) diagnosis requires integrating neuroimaging with heterogeneous clinical evidence and reasoning under established criteria, yet most multimodal models remain opaque and weakly guideline-aligned. We present AD-Reasoning, a multimodal framework that couples structural MRI with six clinical modalities and a rule-based verifier to generate structured, NIA-AA-consistent diagnoses. AD-Reasoning combines modality-specific encoders, bidirectional cross-attention fusion, and reinforcement fine-tuning with verifiable rewards that enforce output format, guideline evidence coverage, and reasoning--decision consistency. We also release AD-MultiSense, a 10,378-visit multimodal QA dataset with guideline-validated rationales built from ADNI/AIBL. On AD-MultiSense, AD-Reasoning achieves state-of-the-art diagnostic accuracy and produces structured rationales that improve transparency over recent baselines, while providing transparent rationales.
Paper Structure (12 sections, 8 equations, 3 figures, 3 tables)

This paper contains 12 sections, 8 equations, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. AD-MultiSense Dataset
  4. AD-Reasoning Framework
  5. Model Architecture
  6. Multimodal Fusion Layer (MFL)
  7. Large Language Model Integration
  8. Training Strategy
  9. Experiments
  10. Quantitative Analysis
  11. Qualitative Analysis and Ablation Study
  12. Conclusion

Figures (3)

  • Figure 1: AD-Reasoning framework. Pretraining aligns sMRI and clinical data representations via encoders, SFT tunes LLMs using diagnostic rationales and RFT optimizes with GRPO for NIA-AA compliant structured outputs.
  • Figure 2: Our AD-MultiSense dataset Construction pipeline: disease-level reports are generated via evidence-augmented reasoning under clinical guidelines with self-refinement for diagnostic validity.
  • Figure 3: Inference example of AD-reasoning.