Our inspiration was in combination of predictive analytics with the real-time mapping of coronavirus that has taken off in recent weeks. We wanted a place for users to be able to track their own data to understand their risk of contracting or spreading the virus.
What it does:
The map uses both pre-existing population movement data and user-submitted information (both anonymized) to reveal current danger spots along with risk assessments for any given location.
How we built it:
We had 3 teams to build our product : User-Interface, Machine Learning, and Databases to structure the core of the proof-of-concept product. The three teams communicated regularly over shared Slack channels, designed our goals in accordance with one another’s needs, and utilized the services available through AWS to generate the backbone of our workflow.
Challenges we ran into
It was difficult to find accessible data that was localized enough to yield novel predictions, while remaining anonymized to protect user identity. Additionally, deciding which predictive model to use and how to concisely integrate the findings into the user-facing product.
Accomplishments that we're proud of
A fully functional platform that will continue to collect user data, improve upon predictions, and allow citizens to stay informed about the safety of their neighborhoods!
What's next for CoronaVision
Build out the predictions and information gathering to all major cities, and eventually all regions, in the U.S. Increase awareness of the product, as well as reinforce efforts to maintain the privacy of data given by users.