<3disease-ml

A web-based machine learning platform targeted at doctors for identifying the precese of heart disease in their patients given some symptoms.

Simple Use

Filling out the form with relevant patient data will return a page contatining the presence of heart disease (Likely or Unlikely). Our nerual network, trained on the 1989 data set collected by Robert Detrano, MD, PhD, is ~99% accurate through simple methods of analysis.

Directory Structure

|
--/templates #html pages
|
-fixdata.py #fills in the missing points in data set
|
-hdnn.py #the predicitive neural network
|
-heart.py #Flask script

Development

To set up the environment for further development the following command can be used:

sudo pip install numpy tensorflow keras Flask jinja2

Notes

• More detailed information can be found in the presentation included in this repo (.pdf file)
• This software is open source and can be used by anyone for any purpose whatsoever
• Created by Sam A. Markelon, Andy Guo, Kyle Fujio, and Julia Lagace at Hack UMass V (Nov. 2017)

Built With

• html
• python