aiqbench analysis
This research introduces a hierarchical Mixture-of-Experts (MoE) framework that effectively fuses frozen imaging models with structured electronic health records to diagnose interstitial lung disease. By employing two-stage gating mechanisms, the architecture achieves superior diagnostic accuracy while maintaining interpretability across anatomical regions and clinical feature groups. The approach demonstrates how specialized expert routing can leverage heterogeneous medical data sources without compromising stable representation learning from pre-trained vision models.
Hierarchical MoE for Multi-Modal ILD Diagnosis
arXiv:2608.25261v1 Announce Type: new Abstract: Mixture-of-experts (MoE) models combine specialized predictors under learned routing, offering a principled mechanism for leveraging heterogeneity in medical data. We present a hierarchical multimodal MoE for interstitial lung disease (ILD) classification that integrates a frozen, pre-trained imaging expert with structured electronic health records (EHR) via two-stage gating. A modality-level gate assigns patient-specific weights to imaging and EHR predictions, while a sub-gating module decomposes the EHR branch into clinically defined feature groups with learned, group-specific contributions. This design preserves stable imaging representations while enabling input-dependent clinical weighting and explicit EHR specialization. Under strict patient-level cross-validation, the model achieved the highest mean AUC among the evaluated methods (0.8750 +- 0.0443), compared with 0.8646 for imaging-only REN and 0.7685 for SwinUNETR. The framework extends interpretability across anatomical regions, imaging--EHR utilization, and clinically defined EHR feature groups.
Authors:Alec K. Peltekian, Gorkem Durak, Halil Ertugrul Aktas, Carrie Lynn Richardson, Mary Carns, Kathleen Aren, GR Scott Budinger, Anthony J. Esposito, Alexander Misharin, Alok Nidhi Choudhary, Ankit Agrawal, Ulas Bagci
Abstract:Mixture-of-experts (MoE) models combine specialized predictors under learned routing, offering a principled mechanism for leveraging heterogeneity in medical data. We present a hierarchical multimodal MoE for interstitial lung disease (ILD) classification that integrates a frozen, pre-trained imaging expert with structured electronic health records (EHR) via two-stage gating. A modality-level gate assigns patient-specific weights to imaging and EHR predictions, while a sub-gating module decomposes the EHR branch into clinically defined feature groups with learned, group-specific contributions. This design preserves stable imaging representations while enabling input-dependent clinical weighting and explicit EHR specialization. Under strict patient-level cross-validation, the model achieved the highest mean AUC among the evaluated methods (0.8750 +- 0.0443), compared with 0.8646 for imaging-only REN and 0.7685 for SwinUNETR. The framework extends interpretability across anatomical regions, imaging--EHR utilization, and clinically defined EHR feature groups.
Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
arXiv:2608.25261 [cs.AI]
(or arXiv:2608.25261v1 [cs.AI] for this version)
Full text under the publisher's open licence · 180 words
Related coverage
A Clinical Point Cloud Paradigm for In-Hospital Mortality Prediction from Multi-Level Incomplete Multimodal EHRs
arXiv · Oct 8, 2026
Benchmarking and Enhancing LLMs for Rule-Intensive Review of National Standard Documents
arXiv · Aug 7, 2026
MARCUS: An agentic, multimodal vision-language model for cardiac diagnosis and management
arXiv · Sep 16, 2026
Multi-modal cross-domain mixed fusion model with dual disentanglement for fault diagnosis under unseen working conditions
arXiv · Jul 21, 2026