Authors

Inspiration - ClosedLoop.ai is healthcare's data science platform. We feel that we are uniquely positioned to help in the fight against COVID-19. About a month ago, we were discussing how we can get involved with COVID-19 with our team of data scientists and healthcare professionals. Being one of the top 25 organizations selected in the CMS AI Challenge (out of hundreds of submissions), we developed predictive models for unplanned hospital admissions, Influenza, and other chronic conditions. With our experience developing these models and having access to CMS data, we knew we could predict whether individuals would experience severe complications should they contract the virus. This is where it all started for us at ClosedLoop, and we were eager to lend a hand.

What it does - The model uses data from CMS to create a vulnerability index. We have already scored over 2 million individuals with the data, and we wanted to build the site to offer this scoring mechanism to the wider population. The individual will select information about their health, age, and gender and we will provide their likelihood for developing severe complications should they contract COVID-19. We have also incorporated a symptoms checker (using CDC guidelines) to provide guidance for the individual on whether they should seek immediate medical attention. Our hope is that individuals can use this information to make their best decision on whether to shelter in place to prevent from contracting the disease, and help loved ones that may be more at risk.

How I built it - We used our models built in python, front-end built with react js.

What's next for C-19 Vulnerability Index - Powered by ClosedLoop.ai - We are currently working on incorporating the CMS Blue Button API to further improve the score. Blue Button API includes additional claims data, drug data, and more where the user can authorize ClosedLoop to use their data to provide the vulnerability score.

Hackathons

Technologies

amazon-web-services, python, react

Devpost Software Identifier

256710