marswen / wardwise

A ward round assistant who helps medical personnel to query medical record data with LLM

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Wardwise

中文 | English

Background

  • In clinical activities such as ward rounds and consultations, medical staff need to frequently query patients' test results and progress.
  • Searching for information from electronic medical records involves reasoning processes such as time conditions, logical conditions, and terminology standardization. Simple structured queries are difficult to meet the requirements.
  • Large language models have changed the task logic of natural language processing and can explore application opportunities under new ideas.

Objectives

  • Establish a workflow framework for electronic medical records and large language model inference.
  • Implement an interactive question-and-answer interface for electronic medical record information.
  • Explore the best practices of using large model inference in different scenarios and domains.

How to use

  • Load Data:
    load_data.py:
    - digest_doc: Extract essential information such as ID and timestamps from plain text based on the minimal information model.
    - load_std_data: Load data in standard fields.
  • Retrieve Medical Records:
    query_record.py: Find medical records that meet the requirements.
  • Answer Questions:
    answer_question.py: Answer query questions based on medical record entries.
  • Example:
    Seeexample.py

About

A ward round assistant who helps medical personnel to query medical record data with LLM

License:GNU Affero General Public License v3.0


Languages

Language:Python 100.0%