Using math and machine learning, researchers predict early flu spike in Canada, U.S.

A team of Canadian and American researchers says their method of forecasting the rate of transmission of infectious diseases is predicting an earlier spike in influenza cases this year.

Using mathematics and machine learning, the researchers analyzed data from late 2015 to September 2024, incorporating weather conditions, policy choices and movement patterns (drawn from cell phones) to predict how diseases like influenza and COVID-19 might spread. 

The team expects there will be more than 1,600 new flu cases confirmed each day in U.S. laboratories by the end of November, nearly double the number of cases during the same time period last year.

“For the prediction in this report, we applied U.S. data, but I think we would state that if we used Canadian data, similar conclusions would be drawn,” said Hao Wang, director of the Interdisciplinary Lab for Mathematical Ecology and Epidemiology at the University of Alberta. 

Wang, who is also a full professor and Tier 1 Canada Research Chair in mathematical biosciences, said the team has predicted the initial surge of flu cases this year will come a month earlier than it did last year. 

He said he hopes this information informs public health officials.

“We hope we can make some impact on the decision making,” he said. 

Wang said he and his colleagues have published multiple papers about their novel forecasting approach. 

Last year, he and Xiunan Wang, an assistant professor in the department of mathematics at the University of Tennessee at Chattanooga, published a paper about it in the peer-reviewed SIAM Journal on Applied Mathematics.

Junling Ma, a full professor in the department of mathematics and statistics at the University of Victoria, who was not involved in the U of A team’s research, said he has not examined influenza patterns in recent years but is familiar with the model and thinks it is suitable for capturing the key features of influenza.

“I think if it happens, their model should be able to capture the feature that it goes up faster and earlier this year, so I would trust their result at this stage” he said.

He said public health officials could prepare for an earlier flu surge by ensuring enough hospital beds are available and encouraging people to get flu shots sooner.

According to the federal government, influenza is among the top 10 leading causes of death in Canada and it causes thousands of hospitalizations every year. 

Last year, less than a quarter of Albertans received the flu shot and the province saw its deadliest flu season in 15 years.

The provincial government announced Monday that flu shot appointments will be available starting October 15.

Health Minister Adriana LaGrange said during a news conference that the government is anticipating a flu season very similar to last year.

“There’s procedures in place, policy in place to ensure that we have the capacity that we need,” she said.

André Gagnon, a spokesperson for the Public Health Agency of Canada, said in a statement that it’s difficult to predict how influenza will behave, but flu season tends to begin in mid-November, peak in late December to February and end by late May.

Gagnon said the public health agency does not use artificial intelligence in its forecasting to predict transmission rates.

“There is currently no indication that influenza cases are increasing earlier than we would expect, but PHAC will continue to monitor and report weekly,” he said.

Ghada Haggag, a pharmacist who owns and manages All Care Pharmacy in Edmonton’s Capilano neighbourhood, said there has been strong demand for flu and COVID-19 shots this year — far more than in 2023.

She said last year people may have been experiencing vaccine fatigue, but many seniors in the area have made two immunization appointments this fall — one for COVID-19 and the flu and the other for respiratory syncytial virus (RSV).

RSV vaccines are given two weeks before or after COVID-19 and flu immunization vaccines.