COVID-19 Datathon Based on Deidentified Governmental Data as an Approach for Solving Policy Challenges, Increasing Trust, and Building a Community: Case Study

TL;DR

This paper describes a government-led COVID-19 datathon in Israel that used deidentified data to develop policy-relevant models, foster community trust, and improve future collaborative data science efforts.

Contribution

It introduces a novel approach to organizing a government-sponsored datathon with deidentified data to address health policy challenges and build trust among stakeholders.

Findings

Increased trust in the Ministry of Health among participants.

Development of potential data science methods for national policies.

Gained insights into organizing effective government-led datathons.

Abstract

Triggered by the COVID-19 crisis, Israel's Ministry of Health (MoH) held a virtual Datathon based on deidentified governmental data. Organized by a multidisciplinary committee, Israel's research community was invited to offer insights to COVID-19 policy challenges. The Datathon was designed to (1) develop operationalizable data-driven models to address COVID-19 health-policy challenges and (2) build a community of researchers from academia, industry, and government and rebuild their trust in the government. Three specific challenges were defined based on their relevance (significance, data availability, and potential to anonymize the data): immunization policies, special needs of the young population, and populations whose rate of compliance with COVID-19 testing is low. The MoH team extracted diverse, reliable, up-to-date, and deidentified governmental datasets for each challenge. Secure remote-access research environments with relevant data science tools were set on Amazon Web. The MoH screened the applicants and accepted around 80 participants, teaming them to balance areas of expertise as well as represent all sectors of the community. One week following the event, anonymous surveys for participants and mentors were distributed to assess overall usefulness and points for improvement. The 48-hour Datathon and pre-event sessions included 18 multidisciplinary teams, mentored by 20 data scientists, 6 epidemiologists, 5 presentation mentors, and 12 judges. The insights developed by the 3 winning teams are currently considered by the MoH as potential data science methods relevant for national policies. The most positive results were increased trust in the MoH and greater readiness to work with the government on these or future projects. Detailed feedback offered concrete lessons for improving the structure and organization of future government-led datathons.