engleski

Understanding COVID-19 pandemic in Croatia using veridical data science

The recent pandemic continues to plague our society, while nations are making tremendous efforts to respond effectively and mitigate risks through targeted actions, from the individual to the collective level. Because these actions have long-term impacts on society, they must be planned and executed with care. To support the global response to novel strains of coronavirus, various relevant data are being collected and shared, providing unique resources to scientific and data driven research communities and policy makers. Understanding and properly interpreting these data is the most important tool in the fight against the pandemic and its all-pervasive impact on our lives. Here, we use openly available governmental, socio-economic, political and public health data and investigate the temporal dynamics of the COVID-19 pandemic in Croatia with the aim of understanding, explaining, visualizing and predicting its nationwide and EU-wide impact under the Croatian policy regime. Motivated by the recently proposed Predictability, Computability and Stability (PCS) framework for Veridical Data Science, we present a new data analysis methodology that answers specific questions about the course of the pandemic in Croatia at the national and subnational levels. We build on and extend the principles of statistics, machine learning, and scientific inquiry, and embed the scientific principles of prediction and reproducibility into data-driven decision making. We estimate the causal impact of the application of health policies, social distancing measures, geographic mobility, and public events on outbreak spread metrics. We provide a comparison across EU countries, an overview of efforts by other approaches, and an assessment of the effectiveness of the methods used. Our work supports well the methodological foundations of Veridical Data Science and trustworthy AI needed to solve problems of great societal importance.

COVID-19 data ; veridical data science ; machine learning ; causal inference ; stringency index ; mobility data

nije evidentirano