Winners identified location-based risks, developed apps to calculate infection risk, and delivered data-driven recommendations for Los Angeles County’s reopening stages.
Data scientists determined to facilitate transitions to reopen the economy participated in a two-week “2020 COVID-19 Computational Challenge” (CCC) in mid-June. The challenge was to “provide guidance for risk mitigation to serve” Los Angeles County. Additionally, the solution “must incorporate the ethical protection of individual data and respect data privacy norms.”
The winning teams revealed location-based COVID-19 exposure at different L.A. communities, developed apps for people to calculate their potential for infection, and delivered applicable data-driven recommendations along with L.A.’s reopening stages, officials said.
Of the 66 data-science teams worldwide (comprising a total of 405 contestants) which entered, six projects were chosen by a panel of judges from the City of Los Angeles, L.A. County Department of Public Health, the L.A. Chamber of Commerce, and academia.
SEE: Coronavirus: Critical IT policies and tools every business needs (TechRepublic Premium)
The event was co-hosted by the Global Association of Research Methods and Data Science (RMDS) Lab, which will compile the winners’ findings to deploy a web-based risk assessment app and an alert system to serve the public. The development is much needed as the coronavirus infection numbers in Los Angeles County and California are skyrocketing this week: As of June 28, L.A. County reported 100,417 confirmed cases and 3,325 deaths, and California reported 217,000 confirmed cases and 5,937 deaths.
In fact, in the very location the challenge was reviewed, the Public Health Department (PHD) reported that 80% of restaurants and bars in Los Angeles County are not following COVID-19 precautions. PHD cited a failure to communicate: Business owners are not communicating new rules to staff and patrons. Basically, in inspections of 2,000 L.A. food services, the most minimal of precautions—wearing a mask and social distancing—are not being followed.
The chosen ones
First place: Team USC-ANRG from the University of Southern California Viterbi School of Engineering—risk estimation using SIR, a simplified color-coded risk level for each community. The risk score corresponds to the probability of a healthy person becoming infected by COVID-19 in the future.
Second place: was shared by three teams (project names are between parentheses)
Team DSO from the USC Department of Data Sciences and Operations (“DSO Infection and Risk Scores”), which predicted where and when the risk of contracting COVID-19 is highest, by predicting the number of new infections in a specific neighborhood on a given day. Basically, it estimates which area/neighborhood presents a higher risk of contracting the coronavirus.
Team RPI Solver from Rensselaer Polytechnic Institute (“Location-based risk score for places in the city of L.A.,” which developed a mathematical model to evaluate the location-based risk in L.A., with data from SafeGraph and open-data portals, and generated risk scores for 30,864 different places, such as grocery and clothing stores, gas stations, and more.
Team Contemporary Li from Zhejiang University (“City of L.A. re-open risk evaluation”)
assessed the risk index of COVID-19 infection in L.A. at different stages of the reopening process by building a multi-indicator evaluation system, as well as proposed epidemic prevention recommendations for the government and communities of L.A.
Best application: Team The Padron Peppers from Grinnell College (COVID-19 activity risk calculator”), which studied socioeconomic disparities across L.A. County neighborhoods and how they may have been/be a factor in the spread of the coronavirus based on positive test results. It evaluates the risk of leaving home, based on a personal-risk profile, a neighborhood-based risk profile, and activity based on risk profile.
Rising star in data science: Team HDMA from the Center for Human Dynamics in the Mobile Age used “population, case rate, death rate, and elderly population to determine risk scores of L.A. County neighborhoods” to make a projection for the total deaths and infections for L.A. County in the coming weeks. The risk scores found can extrapolate if a neighborhood will have an above or below average death and case total in the coming weeks.
The CCC was supported by SafeGraph, Snowflake, UCLA Computational Medicine, Esri, Gartner, Mastercard, and the L.A. County Department of Public Health on open data sources, public health policies, epidemiology, COVID risk scoring examples, data ethics, as well as the business perspective on how to create and utilize this risk score.
Winners received cash prizes of more than $3,000, as well as considerations for internship positions at the City of Los Angeles, UCLA Computational Medicine, and other partner organizations.
In addition, they will have one-on-one mentorship with data executives, a recommendation of a technical report for publishing at Harvard Data Science Review magazine, certificates for winners and contestants who make a complete submission, and an invitation to present at IM Data 2020.