This business claims it built a better COVID-19 forecasting model but can’t get government to bite

Scarsin, based in Markham, Ont., north of Toronto, specializes in creating forecasts for the pharmaceutical industry to predict how different treatments affect patient outcomes in cancer, diabetes and infectious diseases. Its clients include multinational drug makers Bayer, Eli Lilly, Gilead, Jansen, and Pfizer.

“We are one of the top firms in the world at doing this kind of work,” said Minshull. “The global 500 companies that hire us all focus on the exact things that Canada needs in the pandemic today.”

So last spring, when the federal government put out the call for industry to pivot to fighting COVID-19, harnessing the “innovative power of Canadian researchers and businesses,” Scarsin dropped everything to help fight the pandemic.

It assigned a third of its staff, including a PhD who had previous experience modeling a pandemic — the H1N swine flu of 2009 — to come up with a better forecasting model for COVID-19.

Forecast models, based on demographics, testing numbers, positivity rates and more can help leaders choose measures to contain the virus. Modelling can also help predict when hospitals could be overwhelmed with patients.

Scarsin said its software can do localized forecasts for the 92 health regions across the country and consolidate them to create a national model.

A powerful platform

Four experts who reviewed the Scarsin platform for CBC News said they haven’t seen any similar models in Canada.

The company’s COVID-19 program tracks 70 possible parameters, which can be adjusted daily. Neither Ontario nor Ottawa has published a model with a similar design or number of inputs.

The parameters include testing rates, locations people were infected, travel between communities, mask use, household demographics, interventions and more.

It can also forecast how many people will experience mild illness or need hospitalization and separate the results by four different age groups.

It even uses location data released by Apple, Facebook and Google to assess if people are reducing social interactions in line with public health directives.

“It looks like a pretty powerful approach,” said Colin Furness, an infection control epidemiologist at the University of Toronto, “[Scarsin has] a lot of flexibility, adaptability there.”

Scarsin said a key difference between its model and those released by governments is that the company’s system can compare different ways to curb the spread of the virus.

For example, Scarsin said its forecasting model can provide predictions for issues such as how a temporary school closure might affect social interactions and the infection rate compared to the closure of fitness facilities or indoor dining.

The interventions one region may take compared to another to control COVID-19 would depend on the unique demographics of its population and characteristics of its businesses, the company said.

Raywat Deonandan, an epidemiologist at the University of Ottawa, said the ability of Scarsin’s forecasts to be responsive in comparing interventions could help public health leaders make better decisions to better control COVID-19.

“That could have a meaningful impact on the path of the pandemic,” said Deonandan.

Ashleigh Tuite, an infectious disease epidemiologist and mathematical modeler with the University of Toronto, agrees, also suggesting the customizable localized data would be particularly useful.

“I’ve heard from local public health units, you know, there’s this desire to be able to have local level models, and there’s just not the capacity at this point for that.”

She sees it as a ready-to-use solution and would like to see Ontario consider giving Scarsin’s system to health regions across the province.

Little government interest

So far, though, Scarsin has struggled to get health officials to look at its model, starting with Ottawa and the Public Health Agency of Canada.

A manager with PHAC told Scarsin in an email  he would try to get the software checked out by his department “because we have a use for it and don’t have the capacity to do it internally,” but nothing came of the exchange.

Scarsin also applied to a federal government program looking for prototypes to “help combat current and future outbreaks of the novel coronavirus“.

Officials with that program rejected Scarsin’s application, saying its model would be too hard to build and test and potentially too expensive for government departments to license and operate.

It took five months for that rejection to come through.

By then, Scarsin had already built and paid for the system on its own.

Passed over by the province

In May an employee at the Ontario Ministry of Health said in an email to Scarsin that the company’s program “is a great solution.”

It was told to submit the idea to the Ontario Together Fund, a $50 million program to help companies advance ideas or products to battle COVID-19.

Five months later Scarsin had still not received a decision about the application or any feedback.

The company was told the fund was “currently seeking decisions on over 6,000 proposals” that were put forward to the government.

“The process runs like the normal bureaucracy,” said Minshull. “[It’s as though] I’m trying to replace a sidewalk in a subdivision.”

One region is using the software

Scarsin also contacted more than a dozen Ontario health regions and cities, including Toronto, Peel and York.

Only York Region expressed interest, and it began using the Scarsin model in September.

This region north of Toronto includes Markham, Richmond Hill, Vaughan and six other cities with a combined population of about 1.2 million.

“I haven’t come across anything that has the capacity that Scarsin had shown us,” said Katarina Garpenfeldt, the supervisor of advanced planning in the region’s Health Emergency Operation Centre

She said the model helps predict how the pandemic might affect the community and allows health officials to “play with the variables” to see how different interventions could pan out.

“What does it mean if 25 per cent of the students opt in for remote learning, and then all of a sudden that number increases to 40 per cent of students opting in for remote learning? How does that impact the anticipated spread of the disease or the case count?”

A proven prediction

The Ontario government has taken heavy criticism for ignoring the advice of public health experts and relaxing restrictions on businesses to help the economy as infections had been rising for weeks.

Last Friday, record-breaking case counts forced Ontario to abruptly revise its new colour-coded plan for COVID-19 restrictions.

However, ​Scarsin’s modelling found that even Ontario’s most “stringent measures” — the control red zones — will not be enough to slow the spread of the virus and prevent lockdowns.

One of Scarsin’s previous forecasts for Ontario proved to be prescient.

On Oct. 2, in a blog post called “Wave 2 will be late, long and local in Ontario,” the company forecasted that Ontario would have a minimum of 1,540 new cases per day by Remembrance Day.

Ontario hit 1,575 cases on Nov 12.

The COVID-19 Modelling Collaborative, a joint effort of scientists and physicians from the University of Toronto, University Health Network and Sunnybrook Hospital — which informs the group of experts the Ontario government uses to implement health policy — predicted the second wave would peak at 1,000 new cases per day in mid to late October.

Good forecasting saves lives

“I think our lives do depend upon good forecasting,” said Chris Bauch, a professor of applied mathematics at Ontario’s University of Waterloo with a specialty in building complex models about infectious disease transmission.

“The whole idea of flattening the curve was based on mathematical models, and that undoubtedly saved many thousands of lives.”

Bauch, who has built two COVID-19 models himself, said he was surprised how quickly Scarsin developed their system and that it was “a pretty amazing tool.”

Throughout the pandemic, the importance of data and forecasting in fighting infectious disease has become increasingly apparent.

Despite that, experts say there’s not an overarching forecasting strategy.

Instead, Tuite said, forecasting tends to be done by small groups of experts who work together on specific problems, or who may be recruited by governments as advisors as has been the case durIng COVID-19.

“I think people may imagine that each province or each health region, or, you know, the country has this master model that’s reading all of the data that’s being generated and producing forecasts every day that are getting updated,” explained Tuite, “but there isn’t, or there hasn’t been to date, a really unified approach to modelling.”

Canada could lose out

Scarsin said it has spent $1.6 million on its COVID-19 model.

The company offered to deploy the system for the federal government across Canada for less than $2 million.

“We priced that project at a level that I can tell you, we would not have made any money on it,” said Minshull.

Now, Scarsin has started pursuing sales in the U.S. and with private companies to recover its investment.

But it may yet have another option at home.

There’s now a new $10 million federal grant program for researchers to develop forecasting tools.

The federal government set it up because – according to the government website – the pandemic has made clear “that Canada would benefit from additional skilled modelling experts.”

Scarsin may apply.

“Some would say that’s maybe not the smartest decision as a CEO,” said Minshull. “But I can’t help but do it.”