Conversational Disease Diagnosis via External Planner-Controlled Large Language Models

Zhoujian Sun,Cheng Luo,Ziyi Liu,Zhengxing Huang

2024-05-20

Abstract:The development of large language models (LLMs) has brought unprecedented possibilities for artificial intelligence (AI) based medical diagnosis. However, the application perspective of LLMs in real diagnostic scenarios is still unclear because they are not adept at collecting patient data proactively. This study presents a LLM-based diagnostic system that enhances planning capabilities by emulating doctors. Our system involves two external planners to handle planning tasks. The first planner employs a reinforcement learning approach to formulate disease screening questions and conduct initial diagnoses. The second planner uses LLMs to parse medical guidelines and conduct differential diagnoses. By utilizing real patient electronic medical record data, we constructed simulated dialogues between virtual patients and doctors and evaluated the diagnostic abilities of our system. We demonstrated that our system obtained impressive performance in both disease screening and differential diagnoses tasks. This research represents a step towards more seamlessly integrating AI into clinical settings, potentially enhancing the accuracy and accessibility of medical diagnostics.

Computation and Language,Artificial Intelligence

What problem does this paper attempt to address?

The problem this paper attempts to address is: how to utilize large language models (LLMs) to actively collect patient data in real medical diagnostic scenarios and improve the accuracy of disease screening and differential diagnosis by simulating the planning capabilities of doctors.

Specifically, although most existing large language models perform well in disease diagnosis tasks based on existing patient information, they lack the ability to actively collect patient data. This limits their value in actual clinical applications because doctors usually need to collect patient information from scratch, and patients often cannot fully describe their symptoms. Therefore, this study aims to develop a diagnostic system based on large language models that can actively ask patients relevant questions like doctors, thereby gradually collecting the necessary information for diagnosis. The system includes two external planners: the first planner uses reinforcement learning methods to generate disease screening questions, and the second planner utilizes large language models to parse medical guidelines for differential diagnosis. In this way, the researchers hope to enhance the system's planning capabilities, making it more effective in assisting doctors with diagnosis in real medical scenarios.