Asymptomatic Covid tracking (credit card fraud approach)

[mz1]

Github: marcusze/covid-tracking

Background

More than 1/3 of the retirement homes in Sweden report Covid-19, despite having strict visit restrictions and health controls of employees.
The objective of this project is to identify asymptomatic Covid-19 carriers among employees in retirement homes quicker and more accurate.

Solution

We find inspration from fraud detection of skimmed credit cards.
The analogies are

Credit card fraud	Covid tracking
Credit card transactions	Patient visits
Fraudulent transactions	Infected patients
Skimmed terminals	Asymptomatic carriers

Inputs:

1. Visit logs
2. Infected risk patients / elderly

Output(s):

1. Ranking of most probable asymptomatic carriers among caretakers
2. Probability of future infected patients

Screenshots

Check out the detailed project decription, illustrations and screenshots in the shared pdf here.

Illustration of the search algorithm

Data model from simulations

Next steps & Contributions

1. Live trials and feedback: Looking for collaboration partners to conduct real searches on real data. They will be able to customise the solution to their needs.
2. Algorithm and data model improvements: Open for suggestions
3. Tech improvements: Open for suggestions