AI used to track and analyze COVID long haulers

TORONTO – The Vector Institute, an artificial intelligence organization based at the MaRS tech hub in Toronto, has teamed up with telecommunications company Telus Corp., consulting firm Deloitte and diagnostics and pharmaceuticals business Roche Canada to help healthcare professionals learn more about the symptoms that people with a long-lasting form of COVID experience.

They built an artificial intelligence framework that used machine learning to locate and process 460,000 Twitter posts from people with long COVID, the Canadian Press reported. Long COVID is defined by the Canadian government as people who show symptoms of COVID-19 for weeks or months after their initial recovery.

The framework parsed through tweets to determine which are first-person accounts about long COVID and then tallied up the symptoms described. It found fatigue, pain, brain fog, anxiety and headaches were the most common symptoms and that many with long COVID experienced several symptoms at once.

Replicating that research without AI would have taken a huge number of hours from staff members, who would have had to manually locate hundreds of thousands of social-media posts or people and siphon out those without long-COVID or first-person accounts and count symptoms.

“AI is very good at taking large sets of large amounts of data to find patterns,” said Cameron Schuler (pictured), Vector’s chief commercialization officer and vice-president of industry innovation. “It’s for stuff that is way too big for any human to actually be able to hold this in their brain.”

The framework speeds up the research process around a virus that is quickly evolving and still associated with so many unknowns.

So far, long COVID isn’t well understood. There’s no uniform way to diagnose it nor a single treatment to ease or cure it. Information is key to giving patients better outcomes and ensuring hospitals aren’t overwhelmed in the coming years.

A survey conducted in May, 2021, of 1,048 Canadians with long COVID, also known as post-COVID syndrome, found more than 100 symptoms or difficulties with everyday activities.

About 80 percent of adults surveyed by Viral Neuro Exploration, COVID Long Haulers Support Group Canada and Neurological Health Charities Canada reported one or more symptoms between four and 12 weeks after they were first infected.

Sixty percent reported one or more symptoms in the long term. The symptoms were so severe that about 10 percent are unable to return to work in the long term.

Researchers and those behind the technology are hopeful it will quickly contribute to the world’s fight against long COVID, but are already imagining ways they can advance the framework even further or apply it to other situations.

“This is a novel kind of tool,” said Dr. Angela Cheung, a senior physician scientist at the University Health Network, who is running two large studies on long COVID.

“I’m not aware of anyone else having done this and so I think it really may be quite useful going forward in health research.”

Researchers say preliminary uses of the framework show it can help uncover patterns related to symptom frequencies, co-occurrence and distribution over time.

It could also be applied to other health events such as emerging infections or rare diseases or the effects of booster shots on infection.