Researchers will find a consensus between prominent COVID-19 model predictions related to infections and fatalities and create a dashboard to assist governments, corporations, organizations, and the public to evaluate models for better policy and decision making.
In response to the COVID-19 pandemic, decision-makers and the general public have come to rely on a myriad of models to simulate the SARS-CoV-2 virus spread and predict the number of infections and fatalities.
To address the urgent need to compare between the wide-range of existing COVID-19 models and their predictions, the NSF has provided approximately $200,000 to a multi-disciplinary team of George Mason University (Mason) scientists to find a consensus among various model predictions.
Over the next few months, researchers will collect salient COVID-19 models worldwide and cross-examine their assumptions and predictions to find areas in agreement.
The study will undergo two phases—first aligning the models and collecting data on their predicted number of COVID-19 infections and fatalities over a period of time. Next, the team will use a data mining approach that finds clusters among model predictions, indicating agreement.
The results of the clustering analysis will be presented using an interactive web-based dashboard that aims to make the different model assumptions and uncertainty underlying these predictions more transparent.
“Some of these COVID-19 models produce radically different predictions, creating confusion and mistrust over their use,” said Taylor Anderson, PI and Assistant Professor in the Department of Geography and Geoinformation Science in Mason’s College of Science.
The Mason research team will create an interactive COVID-19 Ensemble Dashboard (CED) to display research results online for the public to view starting in late summer, 2020. The CED will serve as an open and accessible tool to inform the public, fellow researchers, and decision-makers where existing models agree and disagree on predictions.
The team will move quickly to deploy the tool. “Model predictions are critical to rapidly develop policy interventions that mitigate COVID-19, to anticipate impacts on health care resources, and to strategize how best to impose and lift public health guidelines,” said Hamdi Kavak, co-PI, Assistant Professor in the Department of Computational and Data Sciences and faculty collaborator in Mason’s Center for Social Complexity.
“Understanding the uncertainty within and across predictions is paramount for effective decision-making to save lives and resources,” said Andreas Züfle, co-PI and Assistant Professor in the Department of Geography and Geoinformation Science.
In addition to Drs. Anderson, Kavak, and Züfle, the Mason research team includes Dr. Joonseok Kim, Post-Doctoral Research Fellow in the Department of Geography and Geoinformation Science.
For more information, contact Taylor Anderson.