What can a bunch of Life Sciences students do without a lab to fight against Covid-19? Our inspiration came from the need for estimation and forecast to balance a lack of medical tests and of tangible ressources. What if we did not need to test everyone or if we had a way to target who should be tested first? How could we use such a tool to end containment safely? We are creating a tool that would allow governments to visualize herd immunity and the need and pertinence of an end of containment by targeted populations.
What it does
We created a survey that collects information about the user on these two themes :
- the user's state of immunity with regards to Covid-19 - we don't know about any total immunity but it could be partial and we assume that people who have already been infected with Covid-19 in the past weeks or months are less likely to have it again
- the part of the user's professional activity that can be carried out in remote work - his/her need to be able to go out to perform this activity
We used the answers to the survey to attribute an individual score to the user that represents whether it would be pertinent to start by him/her for a partial end of containment rather than somebody else. Of course we assumed that if the user is in the at risk population he/she should not go out of containment - this will be determined by the user.
We combined Gaussian distributions with specific mean and deviation to obtain a probability for the user to be in one interval (for the score) or the another.
This tool can be used to plot the data obtained by region of Switzerland, by professional field or by age group - to target a specific population for a progressive end of containment. Also, it could allow to target who should be tested or not for Covid-19 or its specific antibodies.
How I built it
Our survey collects informations about four themes to evaluate the risk associated with the user's "release" of containment. The goal is to evaluate whether it would be pertinent to chose this individual (or population)
personal information: The person’s Canton and whether or not he or she is working from home is additionally asked to evaluate the contamination risk.
Covid-19 testing: This is whether or not the individual has done a Covid-19 test or has been tested positive by a Doctor. It leads to evaluating whether the person is immune or still sick.
household and hygienic measure: Evaluation of the individual’s sanitary actions1 (hand washing, laundry, sneezing and coughing in elbow, and hugging/shaking hands), the possible Covid-19 specific symptoms and whether or not he or she is confined alone.
possible risky social contacts outside the household: Those include public transports, grocery shopping and physical contacts with possibly infected people.
The survey is a Google form and answers to these questions are collected with Matlab. Then a Matlab program measures the final score.
In order to determine the score, 10 categories are chosen, ranging from 0-10: “No societal risk for un-containment” to 90-100: “Person that must not be uncontained”. All survey questions will have a number of answers and a specific standard deviation. For each answer an average risk score is defined, following a Gaussian probability curve, each category will receive a new score.
Challenges I ran into
We are not experienced programmers so the algorithm is still an ongoing project as well as the forecasting of how probabilities will impact each other. Besides, we are working on a very topical issue so we might lack of hindsight.Then it is difficult to obtain reliable data.
Accomplishments that I'm proud of
We worked and tried to finish this project to the best of our abilities and we are very proud to have come this far in our reflection.
What I learned
We learned to work as a team (even if only on Zoom) and design a project from scratch. We use different resources available search as coding on Matlab, using mathematic tools to estimate probability distributions, searching for reliable data and relevant sources.
We also planned to create a platform on which the questionnaire would be available so that people could make changes about their current situation. Unfortunately, the creation of a site required a license or a certain cost, so we decided not to use the HTML language and work with google form.
We hope that our project could be used as a tool for un-containment. It could thus allow a regulated use of the tests, limiting the risk of new contamination while ensuring the economic need for un containment of the region of Switzerland.
In the future we could create a better interface and deepen our questions based on the results obtained, maybe some machine learning could help to correct the model.