Smart Screening Tool

Inspiration

The COVID-19 crisis has created an unprecedented situation in terms of the sudden impact in the economy. Protecting public health is the current primary issue for all EU countries. That said, policy makers need to find a balance between protecting the population against the pandemic and protecting the economy to limit the negative effects of this pandemic. Within that context, the return of the active population to their work is a primary issue and has to be addressed as soon as possible. As a solution to this challenge, EU countries are evaluating several de-confinement strategies that are primarily based on the broad use of COVID-19 tests.

The objective of the project is to roll-out a generalized, scalable, yet resource-optimised strategy for de-confinement with a combination of COVID-19 testing and surveying based on machine learning blockchain, and artificial intelligence techniques. The result would be a risk-based approach for targeted COVID-19 testing that could be used in any de-confinement situation.

Innovative technologies leveraging data science for risk assessment, machine learning. artificial intelligence, and blockchain as the ones proposed in this project have been already deployed on other domains in production at large scale to enable decision-makers to better manage risk and optimally allocate resources. For instance, in the banking domain, in the context of consumer credit, credit scoring using machine learning algorithms is already enabling financial institutions to calculate risks of credit application based on the profiles of credit applicants. In the context of Covid-19, the project aims at building a risk model to calculate the risk of Covid-19 positiveness for any citizen, provided his/her answers to a questionnaire.

What it does

Allows individuals to fill-in an electronic questionnaire with health and demographic information, securing and protecting (anonymised) personal data using Compellio’s blockchain technology.

Predicts the likelihood of positive Covid-19 diagnosis of an individual at a given point in time using machine learning (ML) and artificial intelligence (AI).

Enables policymakers to implement well-informed Covid-19 exit strategies and targeted testing of high-risk individuals.

How we built it

Phase 1: Data collection and modelling

a) Design the questionnaire b) Collect anonymised medical and demographic data of a person when that person takes a test for Covid-19 using the questionnaire from step 1.a and securing data quality with blockchain in a GDPR compliant manner c) Link the test’s outcome (Covid-19 positive or negative) to the data collected in step 1.b d) Build a machine learning model on data from step 1.c

Phase 2: Deployment and general availability

a) Use the model from step 1.d to generate a prediction of Covid-19 positiveness of any person b) Target Covid-19 tests for persons having a high likelihood of positive Covid-19 diagnosis c) Link the test result (Covid-19 positive or negative) to the prediction calculated in step 2.a d) Monitor model performance and fine-tune the machine learning model built in step 1.d

What's next

We are incorporated and ready to quickly roll out the system into production with the support of EU member states, health agencies, laboratories, corporates, and investors.

Interested in joining our efforts? Fill in the partnership form on our website: https://covid19smartscreeningtool.launchaco.com/

Built With

• artificial-neural-networks
• blockchain
• compellio-blockchain-registry
• decision-trees
• deep-learning
• ebsi
• gradient-boosting-trees
• logistic-regression
• matplotlib
• numpy
• pandas
• python
• random-forecasts
• random-forests
• scikit-learn
• scipy
• seaborn
• support-vector-machine
• xgboost
• xgboost-montela-machine-learning/ai:-gradient-boosting-trees