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.

rainiera/health-hack