Big Bang of Brothers| Human Factors
Awards & Nominations
Big Bang of Brothers has received the following awards and nominations. Way to go!
Human Factors
The dance of the wolves and the rabbits
Summary
Management of satellite resources from NASA and commercial entities, with the aim of understanding and predicting the dynamics between the mobility indicators of society and the indicators of increase in the cases of COVID-19, and finally integrating them into a unified framework: the predator-prey model.
How We Addressed This Challenge
INTRODUCTION
During part of World War I, fishing was banned in the Adriatic Sea. As human pressure on the trophic chain of this ecosystem decreased, the number and diversity of species changed markedly. This motivated the Italian mathematician Vito Volterra to study this phenomenon, who, together with the American researcher Alfred J Lotka, developed the Predator-Prey model (1). NASA in 2014 partially subsidized a publication, which from our perspective is excellent. In this manuscript, the predator-prey model was applied to evaluate possible collapse scenarios in the carrying capacity of the planet, considering humans as predators and the planet as prey (2). In our project we propose a reverse approach. Humanity is currently facing a new challenge, COVID-19. The impact generated by this virus transcends strictly medical areas, the consequences of which extend to various areas of knowledge, such as the social, psychological and economic sciences. Therefore, it is valid to question how vulnerable we are as a species. In this context, we propose to consider the predator-prey model, defining the mobility of the population as predator and the uninfected people as prey.
APPROACH
In this challenge, the main objective is to find patterns between human activity and cases of COVID-19. These activities may be related to the number of people who go for a walk in a park, or to do the daily shopping, or who simply need to go to work to feed their family. That is why we propose a double approach. Our first objective is to explore and combine the available satellite resources from NASA, Google and Apple to understand if there is a correlation between population mobility patterns and the spread of the virus. Then, based on this initial information, knowing how population mobility patterns influence the encounter rate between infected and uninfected people, we propose a unified approach using the predator-prey model.
How We Developed This Project
ARGENTINA AS A CASE STUDY
In the Argentine Republic a preventive policy of social isolation of five phases was applied.
- F1: The first phase started on March 20 and established strict isolation. In this phase, only essential services were provided and the mobility of the population allowed was up to 10%.
- F2: The second phase started on April 13 and established a managed isolation. Special circulation permits were administered and the mobility of the population allowed was up to 25%.
- F3: The third phase began on April 26 and geographic segmentation was established. Provincial exceptions were generated and the mobility of the allowed population was up to 50%.
- F4: The fourth phase started on May 10 and established a progressive reopening. An administered provincial reopening was authorized and the mobility of the population allowed was up to 75%.
- F5: Phase five was scheduled for May 24 and consisted of establishing a "new normal", where hygiene and care habits were adopted and the mobility of more than 75% of the population was allowed. However, the number of cases detected was higher than expected, so this phase could not be implemented.
This motivated us to ask ourselves if there is a correlation between the mobility of society and the positive cases of COVID-19. Our first step was to detect the number of positive cases in Argentina. For this we use a free tool, NASA SEDAC (3). However, by including the number of positive cases as an analysis variable, biased results can be obtained, due to their cumulative nature. For this reason, we decided to calculate the daily growth rate (DGR), calculated as: No. of positive cases on day * 100 / No. of positive cases accumulated up to that day. Another possible bias when analyzing the data is that the number of positive cases is closely related to the number of tests that are performed. That is why we also include as a variable the percentage of positive tests per day (PPT), calculated as: No. of confirmed cases per day * 100 / Total tests performed per day. We then use the R programming language to access Apple's databases and determine changes in the number of drivers and the number of walkers over time and Google's databases to determine changes in: retail sales and recreation, groceries and pharmacies, activity in parks, activity in transit stations, activity in workplaces and activity in residences (4,5,6,7,8,9,10,11). But again we consider a new problem. Incubation of the virus is approximately 5.2 days (12), therefore: is it correct to compare mobility levels and indicators of positive cases for the same day? We went a little further, and not only did we correlate the DGR and PPT with the mobility indicators, but we also considered the incubation window period (i) of 5 days, and we correlated the indicators of positive cases with mobility indicators of 5 days (that is, the cases detected today probably became infected approximately 5 days ago). Finally, we use Pearson correlations to compare the indicators.
Furthermore, it would be amazing to be able to include the analysis of the data obtained from the image processing of the black marble layer in NASA's WorldView (13). This tool displays visible light emanating from anthropological sources such as city lights and other human-powered patterns. In this way we could include it as a new activity indicator and correlate it with the indicators of confirmed cases. However, we did not obtain access to the images corresponding to the last years, perhaps due to our lack of experience or because they are not available in this region (Latin America and the Caribbean). The available data were in the time range between 2012 and 2016.
THE PREDATOR-PREY MODEL
This model defines a pair of nonlinear first-order differential equations, which are used to describe the dynamics of biological systems in which two species interact, one as prey and one as predator.
- x’(t) = ɑ*x(t)-β*x(t)*y(t)
- y’(t) = γ*x(t)*y(t)-σ*y(t)
Where in our case we propose the following variables:
- Prey = Uninfected people.
- Predators = Number of infected people in circulation.
- x (t) = Prey as a function of time.
- y (t) = Predators as a function of time.
- ɑ = Birth rate of prey.
- β = Predatory efficiency.
- γ = energy benefit.
- σ = Predator death rate.
NEXT STEPS
As the days pass and we have more data on the interaction between the mobility of the population and the number of infections, we will be able to model these variables with greater precision to include them in the predator-prey model with a specific adjustment for each site. Based on this knowledge, the next step is the development of applications, APIs, libraries, among other automated tools that allow the design of a more dynamic preventive social isolation, which helps to regulate the number of infected without exceeding the load capacity of health units. By establishing regulated preventive social isolation, positive cases of COVID-19 would be controlled, allowing societies greater flexibility in the flow of activities to avoid psychological and socioeconomic collapse. In parallel, the time frame is extended to decrease predatory efficiency (β) through the development of new treatments or perhaps an improvement in the acquired immunization of people.
Project Demo
In our repository you will find:
- The R code that includes:
- Scripts for data collection and organization.
- The scripts to perform the Pearson correlations presented in the project and their respective heat maps.
- The scripts for the plots presented. In addition, there are scripts for you to interactively run these charts and make them easier to navigate.
- The spreadsheet that contains the data from which we perform the analysis. (.xlsx)
- The graphics presented.
- And as a bonus track: a beautiful collage of Noe.
- Alternative slides in .pdf.
Project Code
Data & Resources
1. Berryman, A.A., 1992. The Orgins and Evolution of Predator-Prey Theory. Ecology 73, 1530–1535. https://doi.org/10.2307/1940005
2. Motesharrei, S., Rivas, J., Kalnay, E., 2014. Human and nature dynamics (HANDY): Modeling inequality and use of resources in the collapse or sustainability of societies. Ecol. Econ. 101, 90–102. https://doi.org/10.1016/j.ecolecon.2014.02.014
3. https://sedac.ciesin.columbia.edu/mapping/popest/covid-19/
4. R Core Team (2020). R: A language and environment for statistical computing. Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.
5. Joachim Gassen (2020). tidycovid19: Download, Tidy and Visualize Covid-19 Related Data. R package version 0.0.0.9000.
6. Philipp Schauberger and Alexander Walker (2019). openxlsx: Read, Write and Edit xlsx Files. R package version 4.1.4. https://CRAN.R-project.org/package=openxlsx.
7. Alboukadel Kassambara (2019). ggcorrplot: Visualization of a Correlation Matrix using 'ggplot2'. R package version 0.1.3 https://CRAN.R-project.org/package=ggcorrplot.
8. Hadley Wickham and Jennifer Bryan (2019). readxl: Read Excel Files. R packagev version 1.3.1. https://CRAN.R-project.org/package=readxl.
9. H. Wickham. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York, 2016.
10. Claus O. Wilke (2019). cowplot: Streamlined Plot Theme and Plot Annotations for 'ggplot2'. R package version 1.0.0. https://CRAN.R-project.org/package=cowplot.
11. Johns Hopkins Center for Systems Science and Engineering (2020).
12. https://www.worldometers.info/coronavirus/coronavirus-incubation-period/