Grounding AI Explanations in Experience: A Reflective Cognitive Architecture for Clinical Decision Support

2025-09-25

AI Summary

Clinical disease prediction demands both high accuracy and interpretability—a dual requirement unmet by existing ML/LLM approaches: discriminative models achieve accuracy at the expense of transparency, while generative explanation methods lack statistical grounding and logical rigor. To bridge this gap, we propose the Reflective Cognitive Architecture (RCA), a novel multi-LLM collaborative framework for experiential learning. RCA integrates iterative rule refinement, distribution-aware validation, and prediction-feedback-driven self-reflection to enable deep logical modeling of clinical data. Evaluated on three real-world medical datasets, RCA consistently outperforms 22 baseline models, achieving up to a 40% absolute improvement in accuracy. Crucially, it generates explanations that are clinically comprehensible, logically coherent, statistically grounded, and empirically justified—thereby reconciling predictive performance with trustworthy, human-interpretable reasoning.

Technology Category

• Machine Learning: Transparent, Interpretable, Explainable ML
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Effective disease prediction in modern healthcare demands the twin goals of high accuracy and transparent, clinically meaningful explanations. Existing machine learning and large language model (LLM) based approaches often struggle to balance these goals. Many models yield accurate but unclear statistical outputs, while others generate fluent but statistically unsupported narratives, often undermining both the validity of the explanation and the predictive accuracy itself. This shortcoming comes from a shallow interaction with the data, preventing the development of a deep, detailed understanding similar to a human expert's. We argue that high accuracy and high-quality explanations are not separate objectives but are mutually reinforcing outcomes of a model that develops a deep, direct understanding of the data. To achieve this, we propose the Reflective Cognitive Architecture (RCA), a novel framework that coordinates multiple LLMs to learn from direct experience. RCA features an iterative rule refinement mechanism that improves its logic from prediction errors and a distribution-aware rules check mechanism that bases its reasoning in the dataset's global statistics. By using predictive accuracy as a signal to drive deeper comprehension, RCA builds a strong internal model of the data. We evaluated RCA on one private and two public datasets against 22 baselines. The results demonstrate that RCA not only achieves state-of-the-art accuracy and robustness with a relative improvement of up to 40% over the baseline but, more importantly, leverages this deep understanding to excel in generating explanations that are clear, logical, evidence-based, and balanced, highlighting its potential for creating genuinely trustworthy clinical decision support systems. The code is available at https://github.com/ssssszj/RCA.

Problem

• Balancing high accuracy with transparent clinical explanations in disease prediction
• Addressing shallow data interaction preventing deep expert-like understanding
• Overcoming trade-off between statistical accuracy and meaningful explanatory narratives

Innovation

• Reflective Cognitive Architecture coordinating multiple LLMs
• Iterative rule refinement mechanism learning from errors
• Distribution-aware rules check using global statistics