MedSyn: Enhancing Diagnostics with Human-AI Collaboration

2025-05-07

AI Summary

Clinical decision-making is frequently hindered by cognitive biases, incomplete information, and case ambiguity; meanwhile, prevailing large language models (LLMs) rely on static, single-step reasoning, limiting their applicability to dynamic real-world clinical judgment. To address this, we propose MedSyn—a human-AI collaborative diagnostic framework featuring a novel dynamic bidirectional dialogue mechanism aligned with clinical workflows. MedSyn enables iterative, real-time interaction between clinicians and open-source LLMs, supporting on-the-fly questioning, counter-prompting, and progressive refinement of diagnoses and treatment plans. Methodologically, it integrates prompt engineering, clinical decision logic modeling, and simulation-based interactive evaluation. Experiments demonstrate that MedSyn significantly improves diagnostic consistency (+28.6%) and plan completeness (+34.1%), validating the feasibility of open-source LLMs as trustworthy clinical assistants. This work establishes a new paradigm for explainable, controllable AI-augmented clinical decision support.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Humans and AI: Planning and Decision Support for Human-Machine Teams
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Abstract

Clinical decision-making is inherently complex, often influenced by cognitive biases, incomplete information, and case ambiguity. Large Language Models (LLMs) have shown promise as tools for supporting clinical decision-making, yet their typical one-shot or limited-interaction usage may overlook the complexities of real-world medical practice. In this work, we propose a hybrid human-AI framework, MedSyn, where physicians and LLMs engage in multi-step, interactive dialogues to refine diagnoses and treatment decisions. Unlike static decision-support tools, MedSyn enables dynamic exchanges, allowing physicians to challenge LLM suggestions while the LLM highlights alternative perspectives. Through simulated physician-LLM interactions, we assess the potential of open-source LLMs as physician assistants. Results show open-source LLMs are promising as physician assistants in the real world. Future work will involve real physician interactions to further validate MedSyn's usefulness in diagnostic accuracy and patient outcomes.

Problem

• Addressing cognitive biases in clinical decision-making
• Enhancing AI-physician interaction for diagnosis refinement
• Evaluating open-source LLMs as dynamic diagnostic assistants

Innovation

• Hybrid human-AI framework for clinical decisions
• Multi-step interactive physician-LLM dialogues
• Dynamic exchanges with alternative perspectives