⚕️MedDocChat is a chatbot that allows you to converse with medical documents.
What ⚕️MedDocChat can do:
✅ Answer questions about medical documents (in .txt files) in an intuitive chatbot interface
✅ Provides source attribution 💡 from medical document, ensuring trust in chatbot
✅ Stores medical documents in NoSQL distributed database (using MongoDB at the moment)
✅ Maintains conversational history when chatting with document, allowing for context to be maintained
🛠️ Project's Tech Stack:
- Langchain
- FastAPI
- Next.Js
- MongoDB
- OpenAI
-
Ensure that you have Node.js installed, which can be found on the official website
-
Clone this project onto your local machine
git clone https://github.com/will-pang/medical-document-chat
- (Recommended) Using your terminal,
cdinto the project root folder, create a virtual environment:
python3 -m venv venv
source venv/bin/activate
Note that the latest Python version is required (as of writing, it is Python 3.13.2)
- Install all the required Python packages
pip install -r requirements.txt
- Install all the required Node.js packages
npm install
- Create a
.env.localfile in the project root, with the following variables. You can also find an example underdocs\.env.example:
MONGO_URI="YOUR_MONGODB_URI"
DB="YOUR_MONGO_DATABASE_NAME"
COLLECTION="YOUR_MONGO_COLLECTION_NAME"
OPENAI_API_KEY="YOUR_OPENAI_API_KEY"
CONVERSATIONAL_HISTORY_COLLECTION="YOUR_MONGODB_CONVERSATIONAL_HISTORY_COLLECTION_NAME"
- In your IDE of choice, run the development server via the terminal:
npm run dev -- -p 3000
Open http://localhost:3000 with your browser to see the result.
- In a separate terminal, run fastapi using Uvicorn:
uvicorn api.main:app --host 0.0.0.0 --port 8000 --reload
- Once everything is setup, you should be greeted by this landing page:
You can generate examples of medical documents on ChatGPT, or download some examples under docs and upload it (must be .txt format) by pressing the "Upload .txt File" button.
Note: Currently, the "I'm Feeling Lucky" button is not supported. The idea was to pick a random example stored in the database of choice, but I did not have enough time to implement this.
- Once the document is uploaded, you will be routed to a new page that displays the uploaded text, as well as a chatbot where you can start asking questions!
