Streamlit UI

This is MiADE project's prototype user interface for our automatic point-of-care clinical coding work. We are working to evaluate how a user interface should be designed to enable users to take advantage of language processing tools adding structure to data as they write unstructured notes.

Important

This interface is by default deployed with miade-llm, which requires setting API key environment varibles for the LLM providers. Docker Compose will load them from lib/miade-llm/src/.env. Remeber to put your API keys in that file.

Downloading

After cloning this repository, install the submodules with:

git submodule init && git submodule update

Building

Tip

These instructions use the uv package manager, install uv: https://github.com/astral-sh/uv.

Build the custom component with npm:

npm install --prefix src/frontend
npm run --prefix src/frontend build

Running

Run miade-llm:

cd lib/miade-llm/src
source .env && uv run langchain serve

Run the UI:

uv run streamlit run src/main.py -- -b http://localhost:8000

The data directory, where it saves documents, defaults to data in your working directory. -b is the --backend option; langserve defaults to running on port 8000.

To run with a custom data directory:

uv run streamlit run src/main.py -- -d <PATH/TO/CUSTOM/DIRECTORY> -b http://localhost:8000

This will serve the UI on port 8501.

Docker

To deploy with docker compose:

docker-compose up

This will serve the UI on port 5000.