What if we can predict a Covid-19 outbreak? By the time testing takes place, it is too late. The virus would have already transmitted to several different locations by the time we identify a location that had an outbreak. So by predicting an outbreak early can help us contain the virus. Identifying the potential discrepancy between reported cases and the true disease carriers can solve more than half of the problems.

What it does

Our prediction model takes Google search queries into account. Consider a state with ten cities. Out of those ten, there is one city in which there is a sudden surge of Google search queries regarding the symptoms of Covid-19. When the symptoms first start to appear, people don't rush directly to the doctor. They tend to diagnose those symptoms on their own, and most of the time, they do it by searching on Google. So, if there is a sudden surge in the frequency of Google search query regarding Covid-19, or symptoms of Covid-19 in a particular locality, we can know that the population of that area is suffering from the same symptoms, which is most likely Covid-19. So our prediction model finds such an abnormal increase in Google search queries regarding Covid-19 in a particular locality and predicts whether there is an outbreak or not.

How we built it

We build it with HTML, CSS and JS for frontend and used Dialogflow service by Google cloud to provide our end users a simplistic real and easy assistance. We also used google trends data to feed real time input to our AI which it will use to approximate next massive covid breakout.

Challenges we ran into

We all are more or less beginners for hackathons. We definitely stuck at many different instances for example I didn't get GCP credits to implement my chatbot. We also had trouble in hosting the website since it was using a locally created database. We had trouble integrating python application backend with frontend UI, hence the project is still in pieces in the repo.

Accomplishments that we're proud of

Since there were not many nicely documented Google Trends API, we used web-scrapping and managed to build an unfinished API for our specific usage. We were also able to build a covid calculator, which asks you about the symptoms that you have and accordingly spits out the probability of you being symptomatic.

What we learned

We learned that we should start hacking right away, rather give enough time for brainstorming the idea. We aslo learned how to collaborate virtually on a project for which git was very useful.

What's next for UnCovid

We have layed out next steps that we would be integrating into UnCovid platform and make it public, open-source and free for all.

Please do check out the video presentation.

Built With

Share this project: