Skip to content

Latest commit

 

History

History
94 lines (59 loc) · 3.19 KB

README_en.md

File metadata and controls

94 lines (59 loc) · 3.19 KB

[中文版] [English]

Medical Record Structuring Tool (Under Continuous Update)

This tool is a versatile structuring tool that allows fine-tuning common open-source models for various text data processing and analysis tasks. It currently provides integrated functionalities for training, prediction, and evaluation, with the training and prediction components utilizing [llmtuner] as a core package.

It offers the following common structuring types applicable to various scenarios, such as medical case structuring:

  • Single selection

Single selection example

  • Multiple selection

Multiple selection example

  • Extraction

Extraction example

Installation

First, clone this project to your local computer:

git clone https://github.com/JuneYaooo/llm_structure_tool.git

Conda Installation (Recommended)

Method 1
cd llm_structure_tool
conda env create -f environment.yml
Method 2
conda create -n llm_structure python=3.9
pip install -r requirements.txt

Activate the newly created environment:

conda activate llm_structure

Then run the frontend demo:

python app.py

Usage

The structuring tool provides a simple interactive interface in the terminal. You can enter relevant information and select the desired functionality as prompted.

Single Sentence Testing

Enter a paragraph, set the rules, and perform single selection, multiple selection, or extraction.

Example:

Field Type: 提取

Field Name: 肾上腺肿物大小

Original Text: CT检查示左肾上腺区见大小约5.5 cm×5.7 cm不均匀低密度肿块，边界清楚，增强扫描实性成分中度强化，内见无强化低密度，静脉期明显强化。CT诊断：考虑左肾上腺区肿瘤。B超检查示左肾上腺区见4.6 cm×4.2 cm的低回声区，边界清，有包膜，提示左肾上腺实质性占位声像。

Entering an unrelated field, such as "Gastric Tumor Size," will result in "Not mentioned." Extraction example - Comparison 1

Entering a related field, such as "Adrenal Tumor Size," will result in "Approximately 5.5 cm × 5.7 cm." Extraction example - Comparison 2

Training

To be filled

Prediction

To be filled

Evaluation

To be filled

Acknowledgments

  • PULSE: This project uses the PULSE model (a medical open-source large language model from the Shanghai Artificial Intelligence Laboratory).
  • llmtuner: The training and prediction code for this project is based on llmtuner.

Contribution

If you are interested in this project, you are welcome to contribute your code and improvement suggestions. You can participate in the following ways:

  1. Submit issues and suggestions to the Issue page of this project.
  2. Fork this project and submit your improvement suggestions. We will review and merge appropriate changes.