Project Blue Spider

• 

  Blue Spider

• 

  ANN Accuracy Plot

• 

  ANN Loss Plot

• 

  REST API Demo

Inspiration

Considering a progressive decontainement of a population, we assume the individuals who are the less likely to act as a propagation vector for the Covid-19, would be the first to be cleared for decontainement. We therefore can formulate the following problematic: "How can we determine who are the individuals who has the best chances of reintegration in the active life, without presenting a threat to the population by propagating the Sars-cov2 virus?"

What it does

We have crafted a prototype of a Deep Learning solution that aims to forecast individuals who are most at risk of acting as a vector for the propagation of the Covid-19.

How we built it

The prototype is coded in Python and uses Keras library with a Tensorflow backend, a trained model is delivered in the repository based on a dummy dataset. We also deliver a web application exposing a REST API, and using the trained ANN to ease the integration with other software.

Challenges we ran into

Access to personnal data is still a challenge we are facing. Assuming we can access to these data, we do not master the legal aspect implied by processing and storing these data. The time is not an ally, as always in a hackathon ;)

Accomplishments that we are proud of

Developing a full stack open source prototype using Deep Learning, that could help make a difference in pandemic conditions.

What we learned

We increased our knowledge in several technologies:

• Python and Django.
• The Deep Learning, more precisely building Neural Networks and optimising these networks to achieve better accuracy.

What's next for Blue Spider

BlueSpider prototype can be used a solid base to predict people who are at risk of spreading the Covid-19 (or any contagious disease in epidemic conditions). We hope the people who can access the relevant data required to train the BlueSpider ANN make a good use of our code to take the best strategic decisions.

Built With

• deeplearning
• keras
• python
• rest
• tensorflow
• threatvector