Original Paper: https://arxiv.org/abs/2401.02034

By: Wei ZhuWenfeng LiXing TianPengfei WangXiaoling WangJin ChenYuanbin WuYuan NiGuotong Xie


Knowledge of the medical decision process, which can be modeled as medical decision trees (MDTs), is critical to build clinical decision support systems. However, the current MDT construction methods rely heavily on time-consuming and laborious manual annotation. In this work, we propose a novel task, Text2MDT, to explore the automatic extraction of MDTs from medical texts such as medical guidelines and textbooks. We normalize the form of the MDT and create an annotated Text-to-MDT dataset in Chinese with the participation of medical experts. We investigate two different methods for the Text2MDT tasks: (a) an end-to-end framework which only relies on a GPT style large language models (LLM) instruction tuning to generate all the node information and tree structures. (b) The pipeline framework which decomposes the Text2MDT task to three subtasks. Experiments on our Text2MDT dataset demonstrate that: (a) the end-to-end method basd on LLMs (7B parameters or larger) show promising results, and successfully outperform the pipeline methods. (b) The chain-of-thought (COT) prompting method \cite{Wei2022ChainOT} can improve the performance of the fine-tuned LLMs on the Text2MDT test set. (c) the lightweight pipelined method based on encoder-based pretrained models can perform comparably with LLMs with model complexity two magnititudes smaller. Our Text2MDT dataset is open-sourced at \url{

this https URL

this https URL

Summary Notes

Enhancing CDSS with Automated Medical Decision Tree Construction

In healthcare, the way clinical decisions are made can greatly influence patient results. Clinical Decision Support Systems (CDSS) are crucial for improving these decisions, often using Medical Decision Trees (MDTs) to provide structured advice.

Yet, building MDTs manually is slow, requires a lot of work, and struggles to keep up with new medical findings, which limits the quick updating of CDSS.

This post explores a groundbreaking approach to overcome this challenge: automating the creation of MDTs from medical texts through a process known as Text2MDT, a significant advance in medical natural language processing (NLP).

The Problem with Manual MDT Construction

Building MDTs by hand is a detailed task that involves pulling out important medical information from a large amount of literature and clinical guidelines.

This not only takes a lot of time and expertise but also has trouble scaling up, making it hard to keep CDSS current with the newest medical guidelines and research.

The Promise of Text2MDT

Text2MDT marks a major step forward in automating the creation of MDTs from medical texts.

By using the latest in NLP and machine learning, especially large language models (LLMs), Text2MDT aims to automatically turn structured medical knowledge into MDTs, which could transform how CDSS are kept up to date.

Key Features of the Text2MDT Dataset

A crucial part of this process is the development of the Text2MDT dataset, which is annotated by medical professionals and includes information from clinical guidelines and medical textbooks on various medical conditions and treatments.