HeartShake

Logo created using Inkscape (by Iza L)
Diesease, view of page
The prediction algorithm

Inspiration

If you look online, there are many resources that will give you a list of the symptoms of any cardiovascular disease, and how you can prevent them. There are no readily available user-interactive models in which you can answer questions about your activity levels and your personal information in order to know if you are likely to suffer from a cardiovascular disease. We saw this, and we wanted to fix it as best we could. We aren’t doctors, but we did use data provided from doctor’s and what they’ve examined and reported in patients with cardiovascular diseases.

What it does

In our model, we trained a logistic regression model to recognize the common criteria for being diagnosed with cardiovascular disease.

How we built it

We first used the domain that we had, and created a site. We built the website with React and Node.js, as well as the language JavaScript. We then tried to find the dataset that would best fit our idea. After finding our dataset on kaggle - Cardiovascular Disease Dataset - our team trained our model on the data. Each case in the set had 10 attributes, and the entire set had 70000 cases. If a user provides their own information for each of these attributes (age, height, weight, etc), we can tell the user which percentage of the 70000 cases shared their own particular attributes and if those attributes were more strongly related to patients with cardiovascular disease or not. The model reached a nice 82.7% accuracy in determining this likeness.

Challenges we ran into

Some challenges we ran into were making sure that the dataset form on the website would allow the user to input their data. When using the site, the numbers line up for each box in the form. Also, the website's margins were difficult to center, and other teammates had difficulty with it.

Accomplishments that we're proud of

We're proud of understanding what our idea was set out to do (machine learning was a concept that many team members were not familiar with), as well as trying our best to implement all the things were learned throughout the Hackathon. For example, a team member who worked on the logo had limited knowledge in Graphic Design, and was proud to be able to understand what goes into making a logo.