GSEM: Graph-based Self-Evolving Memory for Experience Augmented Clinical Reasoning

2026-03-23

AI Summary

This work addresses the challenge of unreliable experience reuse in clinical decision-making agents, where conventional memory mechanisms introduce noise due to the absence of explicit relational structures, often degrading performance. To overcome this limitation, we propose the first graph-structured, self-evolving memory mechanism tailored for clinical reasoning. Our approach constructs a two-layer memory graph that captures both intra-decision structural relationships and inter-experience associations. It further integrates relevance-aware retrieval with online feedback-driven calibration of node quality and edge weights. This design significantly enhances the accuracy and robustness of memory retrieval and reuse. Evaluated on MedR-Bench and MedAgentsBench using DeepSeek-V3.2 and Qwen3.5-35B respectively, our method achieves average accuracies of 70.90% and 69.24%, consistently outperforming all existing baselines.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Reasoning under Uncertainty: Graphical Models
• Knowledge Representation and Reasoning: Reasoning with Beliefs

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Agentic search
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

Clinical decision-making agents can benefit from reusing prior decision experience. However, many memory-augmented methods store experiences as independent records without explicit relational structure, which may introduce noisy retrieval, unreliable reuse, and in some cases even hurt performance compared to direct LLM inference. We propose GSEM (Graph-based Self-Evolving Memory), a clinical memory framework that organizes clinical experiences into a dual-layer memory graph, capturing both the decision structure within each experience and the relational dependencies across experiences, and supporting applicability-aware retrieval and online feedback-driven calibration of node quality and edge weights. Across MedR-Bench and MedAgentsBench with two LLM backbones, GSEM achieves the highest average accuracy among all baselines, reaching 70.90\% and 69.24\% with DeepSeek-V3.2 and Qwen3.5-35B, respectively. Code is available at https://github.com/xhan1022/gsem.

Problem

• memory-augmented reasoning
• clinical decision-making
• experience reuse
• relational structure
• noisy retrieval

Innovation

• Graph-based Memory
• Self-Evolving Memory
• Clinical Reasoning
• Experience Augmentation
• Relational Structure