PRS8 IMPLEMENTING AN ASTHMA DATATHON TO LEVERAGE PUBLIC DATA TO INFORM DECISION AND POLICY-MAKERS IN BRAZIL

Abstract

to integrate and analyze public datasets to create interactive vizualizations of asthma-related outcomes, through an online open data analytics competition-datathon. A 2-day design-thinking workshop was organized by GSK in partnership with Eretz.bio (startup incubator) from Hospital Israelita Albert Einstein, to define the datathon challenge, rules, timelines and prizes. The theme was to understand the impact of diagnosis, hospitalizations, costs, mortality, structural, social and environmental factors related to asthma in Brazil. An open online call was made to invite teams to develop their projects in 2 weeks. Each team with 3 members had to use at least one database from the Public Health system (DATASUS). The evaluation criteria included: creativity and innovation in epidemiological analysis; technical capability and methodology; quality of dashboards and its visualization. Each criteria was graded from 0 to 5 by an independent judging panel, and the top 3 scores were the winners. There were 240 participants (80 teams) from 15 Brazilian States. Among them, 85% were male and regarding the background, 45% were data scientist, 24% undergraduate students, 18% healthcare professionals, 11% investigators and 2% designers. Novel analytical dashboards included interactive maps and tables describing trends of asthma hospitalizations per day/month in different regions, days of school missed last year due to asthma in school; comparison of asthma to other chronic diseases like diabetes, demonstrating similar impact on costs and resource use. These dashboards were deemed useful for public health administrators at different levels. There was consistency in the analyses of the top teams which provided insights and deeper understanding of the impact of asthma and its multiple effects. Public healthcare databases are increasing, offering the opportunity to generate evidence for decision makers. Countries with open datasets should consider implementing similar activities to increase understanding of healthcare conditions and support decision making.