This is the django interface for Fight Health Insurance a health insurance appeal bot(FHI).
Currently this has both the frontend (HTML & javascript) along with a bunch of "middle-ware" type logic (forms, database interactions, etc.)
In the future this may be seperated out into different projects.
The ML model is generated using this repo.
FHI requires Python 3.10 or 3.11. You can check the version of Python on your system by typing python. If you have a different version of Python two easy ways to get a different version of Python on your system are Anaconda or UV. Once you have a supported version of Python installed, you'll want to install all of the requirements for the project with pip install -r requirements.txt. You can do this inside a virtualenv or conda env.
If you get django.core.exceptions.AppRegistryNotReady: The translation infrastructure cannot be initialized before the apps registry is ready. Check that you don't make non-lazy gettext calls at import time. make sure you are running Python 3.10 or higher.
The run_local.sh can be used to launch django to run locally.
To really test changes you'll likely want access to a model, one option is using this repo and setting HEALTH_BACKEND_PORT to 8000 and HEALTH_BACKEND_HOST to localhost. Deploying locally requires ~ GPU equivalent to a 3090.
If you don't have a GPU handy the other option is to use an external model. The current one setup by default is octoai & you can get a free API key with enough credits to run locally. You'll want to set the enviornment variable OCTOAI_TOKEN to the value of your API key.
Tests are run through tox. If you dont have tox you can pip or uv install it. The tests are broken up into sync and async. You can run all tests by running tox.
An example of running just one test suite is tox -e py311-django50-sync -- tests/sync/test_selenium_appeal_generation.py
We use black for style checking, and it can auto apply many style fixes so if you get a style error just run black fighthealthinsurance.
We use mypy for type checking.