athenahealth hackathon 2015
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
HealthCruncher_IOS/App/HealthCruncher
models
static
templates
.gitignore
LICENSE
README.md
iOS_screens.pdf
main.py
progress_bar.py
requirements.txt
slide_deck.pdf

README.md

HealthCruncher

conceived at the athenahealth MDPhackathon 2015 (July 25-26, 2015)

Static web app demo

Live web app demo / iOS app screenshots

MDPhackathon slide deck (3 minutes)

MDPhackathon's tweet about HealthCruncher

What is HealthCruncher?

Netflix for predicting insurance customer health outcomes.

HealthCruncher leverages machine learning algoritms [sic] to modernize the way insurance companies calculate premiums

  • A "B2B SaaS" provider for insurance companies.
  • A RESTful API that wraps a machine learning algorithm. Insurance companies can provide simple inputs and get trivially parsable JSON objects.
  • Web app and iOS app linked to the API for the hackathon.

Development

  • Some stuff we use: 🐍 flask, DigitalOcean, scikit-learn, anaconda
  • $ python main.py
  • Test at 0.0.0.0:5000
  • Use a virtualenv if you feel like it.

Deployment

  • $ ssh root@strtup.me
  • $ cd /var/www/healthcruncher
  • $ git pull
  • In main.py, debug should be FALSE!!!
  • $ . venv/bin/activate
  • $ nohup python main.py > /dev/null&

Undeployment

  • $ ssh root@strtup.me
  • $ ps -fA | grep python
  • Find the <PID> of the one that has python main.py
  • $ kill <PID>

To-do

  • HTTPS
  • Better the algorithm (actual data, better validation)
  • Better the API (fix some Flask routing issues)
  • Better form validation (CSRF protection!)
  • Streamline deployment (scripts, etc.)
  • Better domain

Who we are

Rainier Ababao - web dev/deployment/data science
Shaun Mataire - iOS/deployment
Hayley Call - business model/slide deck maestro/web dev
Sarah Gorring - business model/slide deck maestro/web dev

The initial idea

A web app that inputs easily accessible patient data and outputs metrics (like heart disease or diabetes predisposition) based on a machine learning algorithm.

Built with ❤ in Austin, TX