On Dec.
31, BlueDot, a Toronto-based company that uses artificial intelligence to track the spread of infectious diseases, alerted its customers about a cluster of unusual pneumonia cases in Wuhan, China.
Nine days later, the World Health Organization confirmed the discovery of a novel coronavirus, later named COVID-19, in Wuhan.
Today, COVID-19 is a pandemic that has spread to 180 countries, claimed more than 83,000 lives, and triggered a near-global lockdown.
And for the moment, the best solution to contain the spread of the virus is to improve personal hygiene and exercise social distancing.
In the meantime, politicians, scientists, and researchers are teaming up to find systematic ways to fight the virus and care for patients.
And they’re getting some much-needed help from artificial intelligence.
Tracking the Spread of the Virus
BlueDot uses a combination of artificial intelligence and human expertise to track the spread of infectious diseases across the globe.
Its algorithms consolidate and analyze data from sources including news reports, statements from health organizations, commercial flights, and livestock health reports.
Using machine learning and natural language processing, BlueDot peruses the sea of data to find patterns that might hint at the beginning of an infectious outbreak.
The results are then reviewed by a team of experts composed of epidemiologists, doctors, veterinarians, and data scientists, who decide which of the signals need further investigation.
The final report is sent out to BlueDot’s customers, such as governments and businesses.
In addition to hotspots, the AI can also predict the spread of infectious and contagious diseases using flight data and movement patterns.
BlueDot successfully predicted several cities where COVID-19 would first spread, after it surfaced in Wuhan.
Under normal circumstances, BlueDot provides its platform as a commercial application.
But these days, the company is helping governments track the spread of COVID-19.
In the future, AI technologies like BlueDot’s can serve as early warning systems to help governments nip pandemics in the bud.
“BlueDot is humbled and grateful for the opportunity to combine our expertise in infectious diseases, big data analytics, and digital technologies with the efforts of the government of Canada to protect lives and mitigate the impacts of COVID-19 here at home and around the world,” said Dr.
Kamran Khan, infectious disease physician and CEO of BlueDot.
“We are in uncharted territory as a microscopic virus is now disrupting our entire planet.
The COVID-19 pandemic has revealed the need to implement systems that proactively manage infectious disease risks which, in our rapidly changing world, are increasing in frequency, scale and impact.
And it is with enhanced preparedness that we can get ahead of these threats to create a healthier, safer, and more prosperous world.”
Detecting COVID-19 Infection in Medical Imagery
Viral test kits are in short supply, and scientists and researchers have been looking for alternative ways to find COVID-19 infections.
One possible solution is the examination of chest X-rays and CT scans, which are more readily available in hospitals and can show infections caused by COVID-19.
The challenge in using chest imaging in diagnoses is that it’s hard to tell the difference between COVID-19 and other infections such as influenza.
The American College of Radiology (ACR) issued a statement in March, advising against the use of chest CT scans and X-ray as the first-line test of COVID-19.
“Viral testing remains the only specific method of diagnosis,” ACR wrote in its advisory.
The position is also supported by the CDC, which states, “Given the variability in chest imaging findings, chest radiograph or CT alone is not recommended for the diagnosis of COVID-19.”
But AI researchers are hoping that computer vision will help where human vision fails.
Several companies have deployed AI systems to detect COVID-19 cases in X-ray and CT scans.
One recent effort is COVID-Net, an open-source deep-learning system developed by DarwinAI and the University of Waterloo.
Recommended by Our Editors
Alex Wong, chief scientist at DarwinAI, says there are subtle differences between COVID-19 and other infections that radiologists might not notice when examining chest X-rays. “The hope here with COVID-Net is that we can leverage AI (specifically, deep learning) to pick up these subtle visual indicators to better differentiate between COVID-19 and other forms of infections, and unveil these visual indicators to clinicians to boost specificity,” he says.
Deep-learning algorithms are especially good at finding small details in visual data that can go unnoticed to the naked eye.
COVID-Net has been trained on COVIDx, a public database that consists of 16,756 chest X-Rays across 13,645 patient cases from not only COVID-19, but other types of lung infections as well.
The diversity of the data will enable the deep learning model to single out the characteristics that define each type of illness and detect them in new X-ray images.
Wong says that while the model is not production-ready, preliminary results are very promising in differentiating between COVID-19 and other infections.
The model will be improved as more data becomes available.
“We feel strongly that a large-enough sample size would make a big difference in improving COVID-Net as well as develop new deep-learning models for detecting COVID-19 infection,” Wong says.
Nonetheless, Wong stresses that, as advised by CDC and ACR, chest X-rays and CT scans should still be considered as complementary screening tools.
They can be used in facilities where test kits are in short supply or not available.
There are also situations where chest X-rays or CT scans must be done even in a positive diagnosis with viral tests to assess the extent of the infection for treatment and care planning.
“The hope is that the AI can help radiologists to more rapidly and accurately differentiate between COVID-19 infections and other forms of infections (especially important since flus are prevalent still this time of year), and more importantly, reduce the burden for radiologists but enabling other front-line health workers with less expertise to better make diagnosis,” Wong says.
