How Polly began working on COVID
by Kenton White, PhD — Chief Scientist for Advanced Symbolics
In April 2020, when we were first coming to terms with the global pandemic, Canada, through the National Engineering and Scientific Research Council (NSERC), was asking its best researchers to pause their existing research and focus on the ongoing crisis. That’s when Professor Mao of the University of Ottawa approached me about collaborating on a COVID project. I have known Prof. Mao for over a decade, dating back to when I was doing research there that eventually led to the creation of Polly.
Prof. Mao proposed pooling his expertise on developing new mathematical models with our research on measuring and forecasting people’s behavior in order to teach Polly, our AI, how to help in the fight to conquer the virus. Professor Mao is a recognized genius. How could I say no? This was a huge opportunity. Polly could learn some new tricks from a leading researcher, and we could contribute towards Canada’s fight against COVID.
At that time everyone was talking about “the curve” — that smooth, bell shaped graph showing when infections might reach their peak. When we compare the curve against the actual data, it struck us that the real cases were very “noisy” – from a data perspective. They weren’t smooth, instead spiking and dipping and then spiking again. We consulted with several epidemiologists working on the front lines who confirmed our suspicions: having a more accurate forecast of COVID infection rates — a forecast that included the noisy spikes — could help them better fight this disease.
The big idea that ultimately won the NSERC grant competition was adding spatial correlation into the model. Not every region, province and city experiences the curve similarly. Some areas are further ahead; other areas have less intensity. Professor Mao’s group was able to analyze all of the data, learning what leads to higher intensity cases or faster spread through the population. Once these factors are identified, Polly can look for these patterns in the population, identifying regions where she thinks the conditions are just right for a spike in cases, or a rapid spread. After working through the technical details, we were convinced that this could work.
After a stringent evaluation, our project was deemed one of the best in Canada by a panel of experts. We learned in early May that our project was approved! We’ve been working on this research for about 5 months and are still analyzing data for the factors we want to use. Once Professor Mao’s group provides us with the right things to look for, Polly will get to work. Until then, I really appreciate all of the questions and comments — your ideas about what to look for and what potential challenges you see are helpful. Research is about asking the right questions.
Submit a comment
Your email address will not be published. Required fields are marked *