Skip to content

Latest commit

 

History

History
61 lines (38 loc) · 1.71 KB

README.md

File metadata and controls

61 lines (38 loc) · 1.71 KB
Raw

GPT Nurse

This application processes a medical report in PDF format and uses the GPT-4 engine to answer a series of questions based on the report.

Dependencies

The application relies on the following Python libraries:

  • pdfminer
  • openai
  • pandas
  • json
  • os
  • re

Additionally, Docker needs to be installed on your machine to build and run the application inside a Docker container.

Classes

gptNurse

This class performs the following:

  1. Initializes the GPT model and processes the provided medical report.
  2. Extracts text from the PDF report along with its font details.
  3. Converts the extracted PDF content to a dictionary format.
  4. Initializes the GPT model with API credentials.
  5. Uses the GPT model to answer a given question based on the report.

Methods

extract_text_with_font

Extracts text characters from a PDF along with their font name and size.

pdf2dict

Processes the extracted text and groups it into sections based on identified header fonts. This results in a dictionary where each section of the PDF is represented as a key-value pair.

initGPT

Initializes the GPT model using API credentials from a provided JSON configuration.

answer_question

Uses the GPT model to answer a given question based on the medical report. The question is provided in a predefined JSON format.

How to Use

  1. Ensure you have Docker installed on your machine.
  2. Place your medical report in the root directory and name it medical-record.pdf.
  3. Place the task.json configuration in the root directory.
  4. List your questions in questions.txt, placing each question on a new line.

Docker Commands

To build and run the Docker image:

docker build -t gpt-nurse .
docker run gpt-nurse