Humans work fast when faced with a crisis, and COVID-19 has a lit a fire under the science and tech communities. Researchers in England believe a coronavirus cure could be ready by September; scholarly articles are being written, reviewed and published at record speeds; and in a matter of weeks, five million Canadians have comfortably switched to working from home. The world is handling COVID-19 quicker than expected. Intelligent computers, along with their human teachers, play a big part.
Faster and more accurate than any person, AI and machine learning have turned the health world on its head, often doing in seconds what used to take doctors and scientists weeks. That accuracy and efficiency has been invaluable in the fight against COVID-19. Here are some ways Canadian-created AI has helped put the brakes on the pandemic.
How: Being able to speak multiple languages is a tremendous asset in any field. Now imagine you could — in seconds — dissect and interpret more than 100,000 articles per day in 65 different languages. Enter the AI at BlueDot, the Toronto company that on New Year’s Eve, warned its clients of COVID-19, and by January 14, successfully predicted 10 of the first 12 cities to see official cases. The company uses its technology to track more than 150 infectious diseases and syndromes worldwide in near real time. BlueDot’s AI scans official public health sources, moderated health reports, mass media and the world’s flight itinerary, dramatically reducing the time to make a risk assessment, and successfully predicting where a virus could spread.
What’s next: Prime Minister Justin Trudeau announced the government would use BlueDot’s AI platform to track the coronavirus’s spread globally, and inform government policies and decisions. Before the outbreak, BlueDot had no clients in the United States. Since then, several jurisdictions, including the state of California, have applied the company’s services.
How: While researchers around the world have made great gains in the race for a COVID-19 vaccine, no one knows for sure when such an innovation will be approved. Startup Cyclica has taken a different approach, using its machine learning tech to find existing drugs, already approved by the FDA. The best way to understand how Cyclica’s tech works is by analogy. In order to be effective, medications must act like a key, locking and unlocking various proteins to attack a given disease. In the past, that took years of trials and billions of dollars. But President and CEO Naheed Kurji believes existing therapies “can be rapidly deployed to lessen the severity of the disease — a stop-gap until a vaccine is designed.” Sometimes keys can lock and unlock more than one door. In the world of patient care, that represents a massive opportunity.
What’s next: Cyclica has partnered with Beijing institution Materia Medica in the search for existing COVID-19 drugs, with tests currently underway. The startup also launched its Cyclica Stimulus Package: free access to the group’s AI platform for any researcher or biotech company that is either actively fighting the virus, or has seen their work stifled due to the crisis.
How: Like Google for genomic data, health company DNAstack uses machine-learning algorithms to help researchers determine which genetic markers predispose patients to certain diseases; then, by classifying subjects based on their response to medications, treatments can be targeted to specific people and genes. In everyday terms, scientists use DNAstack’s software like a game of global Go Fish, asking the computer questions related to international health data: Does this coronavirus patient appear to have immunity? What region seems to be producing the most devastating cases? How many of the following patients have compatible blood types for my COVID-19 experiment? And so on.
What’s next: DNAstack’s COVID-19 Beacon (compliant with the Beacon API, an international open-source protocol developed by the Global Alliance for Genomic and Health) is helping researchers discover sequences with specific genetic mutations and chart their geographic and evolutionary origins.
How: The most time-consuming and error-prone stage of drug discovery comes when cleaning and structuring data for analysis. In other words, if the information going into the computer is unclear or inconsistent, the computer can’t make sense of it. (For example, incoming Korean data may mark all male patients with an “M,” while parallel German data may denote men with a “1.”) BioSymetrics combines machine-learning frameworks and in vivo experimental research (work within a living organism, such as clinical trials in humans) to remove the fuss and discover new drugs. When pairing experimental methods with its multi-dimensional “barcode” of in vivo phenotypic characteristics, the company helps researchers better target therapies and reduce lab time.
What’s next: BioSymetrics’s predictive method is also proving useful in the coronavirus battle, analyzing emerging patient data sources to help healthcare professionals make better, quicker decisions. And the startup is working closely with its partners to determine if any of its early-stage drugs may be able to mitigate coronavirus symptoms.
How: Headquartered in Toronto and home to the person considered to be the “Godfather of AI,” the Vector Institute is on the leading edge of computer science. Now the organization is using its immense brain power — real and digital — to tackle the concrete problem of COVID-19. The Vector Institute has repurposed 240 new NVIDIA Quadro RTX 6000 graphics processing units, (typically used for best-in-class illustration, animation and deep learning) to analyze Ontario’s public health data and make predictions. This computing power will bring super-fast insights on the risk of spreading the virus to vulnerable people; asymptomatic transmission and outbreaks; allocation of tests, ventilators, and staffing; localized social distancing and closure policies; and, stabilization of supply chains for ventilators and personal protective equipment.
What’s next: “I am deeply proud of the Vector Institute’s community of scientists who have not hesitated to contribute their expertise to combat this pandemic,” said president and CEO Garth Gibson. “This necessary infrastructure will ensure Ontario benefits fully from our high concentration of world-class machine learning practitioners.”
For more information on how MaRS-supported startups are helping in response to the pandemic, visit our COVID-19 directory.