Using machine learning to predict viral transmission rates in Halifax

Halifax, NS (May 26, 2020) Dr. Yigit Aydede, an associate professor of economics in the School of Business at Saint Mary’s University, is leading a study that will analyze the transmission rate of COVID-19 in Nova Scotia.

“Factors that affect the spread of viral infection are multiple and complex,” says Dr. Aydede. “Further, assessing the relative contribution of these factors requires local data because transmission of the same viral pathogen often proves to be different depending on spatial differences.”

Although a more detailed understanding of how viruses transmit could have broad public health implications, Dr. Aydede explains that the lack of data at local levels, until now, has hindered a better understanding about the role that environmental factors play on the transmission of viral pathogens.

“Recent advances in statistics and modeling now make it to possible to reconstruct information from data gathered during outbreaks, allowing for a more refined evaluation,” says Dr. Aydede. “Unlike a causal inference, our empirical objective is to develop predictive models by using advance machine learning methods and identify the importance of predictors.”

Dr. Aydede’s research team is comprised of experts in economics and computer science from Saint Mary’s University, Dalhousie University, and Acadia University. The team hopes to create simulation models that can better calibrate spatial spread in order to inform policy making. A recent grant from the Nova Scotia COVID-19 Health Research Coalition will now help propel their work forward.

Using Nova Scotia COVID-19 test data, Dr. Aydede and his team first plan to analyze the transmission of viral pathogens and the number of positive COVID-19 cases in response to local climatic and air quality conditions. Next, the team will measure local mobility using Apple and Google Application programming interfaces (API). Finally, they will analyze these data with respect to the speed of transmission measured by the demand for the COVID-19 tests, which can be related to the level and the mode of mobility in Halifax. 

“The results could provide invaluable tools to policy makers for predicting when fluctuations would occur and at what level of relaxing the restrictions would not trigger a rise in transmission,” says Dr. Aydede. “With this new understanding, resources could be prioritized to predict and address escalations in demand for health care services and ER visits.”

The team is poised to rapidly deliver the predictive models and their source code to decision makers within a few months.

This research project was funded by the Nova Scotia COVID-19 Health Research Coalition. Partners include the Nova Scotia Health Authority, Dalhousie University, Dalhousie Medical Research Foundation, IWK Health Centre, IWK Foundation, QEII Health Sciences Foundation, Dartmouth General Hospital Foundation, and Research Nova Scotia. The Coalition is dedicated to leading and fostering a research environment that engages our academic partnerships and responds to the current needs of Nova Scotians and our health care system, in addition to maintaining the expertise in innovative research, discovery science, population/social sciences, and health system improvement. This funding partnership provides the opportunity to catalyze COVID-19 related research initiatives and achieve collective social impact. For more information visit