Towards Interpretable Renal Health Decline Forecasting via Multi-LMM Collaborative Reasoning Framework

2025-07-30

AI Summary

Chronic kidney disease (CKD) management faces a trade-off between eGFR prediction accuracy and clinical interpretability, while large multimodal models (LMMs) suffer from high deployment costs, privacy risks, and limited reliability. Method: We propose an open-source collaborative LMM inference framework integrating vision-based knowledge distillation, abductive reasoning, and short-term memory mechanisms to jointly model multimodal clinical data (e.g., medical images and structured biomarkers) and enable logically traceable, guideline-conformant reasoning. Contribution/Results: To our knowledge, this is the first fully open-source system achieving eGFR prediction accuracy comparable to proprietary models (MAE ≤ 2.1 mL/min/1.73m²) alongside high-fidelity clinical explanations—generating nephrology-guideline-aligned inference pathways. Extensive evaluation on real-world data demonstrates robustness, strong generalizability, and clinical utility, establishing a new paradigm for low-cost, trustworthy AI-assisted CKD decision support.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Computer Vision: Multi-modal Vision

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Accurate and interpretable prediction of estimated glomerular filtration rate (eGFR) is essential for managing chronic kidney disease (CKD) and supporting clinical decisions. Recent advances in Large Multimodal Models (LMMs) have shown strong potential in clinical prediction tasks due to their ability to process visual and textual information. However, challenges related to deployment cost, data privacy, and model reliability hinder their adoption. In this study, we propose a collaborative framework that enhances the performance of open-source LMMs for eGFR forecasting while generating clinically meaningful explanations. The framework incorporates visual knowledge transfer, abductive reasoning, and a short-term memory mechanism to enhance prediction accuracy and interpretability. Experimental results show that the proposed framework achieves predictive performance and interpretability comparable to proprietary models. It also provides plausible clinical reasoning processes behind each prediction. Our method sheds new light on building AI systems for healthcare that combine predictive accuracy with clinically grounded interpretability.

Problem

• Predicting eGFR accurately for CKD management
• Overcoming deployment and privacy issues in LMMs
• Enhancing interpretability in clinical AI predictions

Innovation

• Multi-LMM collaborative reasoning framework
• Visual knowledge transfer and abductive reasoning
• Short-term memory mechanism for accuracy