Fellow & Canada CIFAR AI Chair Russ Greiner (Professor, University of Alberta) and a team at the University of Alberta are using machine learning tools and public data to try and predict the future when it comes to coronavirus cases.
As reported in Folio, the team has developed a model that aims to forecast the number of future coronavirus cases that a region might expect. That would give public health officials valuable time to allocate resources and prepare for influxes of patients.
Models that forecast disease cases are nothing new -- -- since the beginning of the pandemic, health agencies have created countless forecasts of the virus’ predicted behaviour. However, most of those models are only accurate in the short term, usually a matter of weeks. The farther into the future one looks, the less accurate the forecasts become.
The University of Alberta team’s approach pulls data from a diverse range of sources: ICU admissions, school closures, commercial restrictions among them. The AI tools built by the team then pours over the data, identifying the data points that are strong predictors of future cases and discarding those that aren’t. With enough data and time, the model can be trained to predict the possible cases an area might see up to ten weeks in the future.
The multidisciplinary team involves students and researchers from a variety of departments at the University of Alberta, including the Department of Mathematics and Statistical Sciences, the Department of Biological Sciences, the Department of Computing Science, and the Faculty of Medicine and Dentistry’s Department of Laboratory Medicine and Pathology.