5 Ways Artificial Intelligence Is Guiding Us Through the Pandemic

August 19, 2020 Health Ideas Staff

Artificial intelligence (AI) has been the stuff of sci-fi theater since long before the Terminator’s Skynet made its debut. But contrary to what Hollywood might have us believe, the real power of AI is not in its ability to replace human intelligence, but to augment it — and in the battle against COVID-19, we need all the help we can get.

From charting the course of the virus to accelerating its diagnosis and treatment, we’re highlighting some of the most promising applications of AI and machine learning in healthcare today.

1. Detecting Potential Outbreaks

AI helped to discover the novel coronavirus in late December 2019, when Toronto-based AI startup BlueDot alerted its government, hospital and airline clients to a cluster of “unusual pneumonia cases” around a Wuhan, China, market. Days later, on January 9, 2020, the World Health Organization released its initial statement on the novel coronavirus.

As COVID-19 cases peak, plateau and decline in different areas at different times, AI systems that “sniff out” signs of outbreaks can enable better decision-making and overall preparedness. For instance, rather than taking a blanket approach to capacity optimization measures like postponing elective procedures or pausing payment integrity, highly accurate AI insights would enable stakeholders to make these decisions on a case-by-case basis — helping to contain costs while protecting the health and safety of those most vulnerable.

2. Enabling Faster, More Accurate Diagnosis

Noting various shortcomings of standard COVID-19 testing methods, Mount Sinai researchers used AI to improve and accelerate the diagnosis of COVID-19-positive patients. As part of the study, published in the journal Nature Medicine, researchers developed an AI algorithm by combining CT chest images with clinical data (symptoms, exposure history and laboratory testing). When applied to a test set of 279 cases, the algorithm demonstrated higher sensitivity than a senior thoracic radiologist (84.3% versus 74.6%, respectively).

A recent report from Johns Hopkins takes the potential of AI a step further, encouraging its use as a means of “diagnosing and assessing coronavirus infection,” rather than merely screening for signs of COVID-19 in the lungs.

3. Remotely Monitoring COVID-19 Symptoms

Northwestern University researchers, in collaboration with Shirley Ryan AbilityLab, developed an AI-enabled wearable device that monitors COVID-19 symptoms. The wireless sensor, which sits at the base of the throat, continuously measures and analyzes coughing and respiratory activity. By picking up on early warning signs, researchers aim to minimize unnecessary hospital visits, instead enabling preventive measures. Currently, the team is using machine learning algorithms to train the device to distinguish between COVID- and non-COVID-like coughs. The project recently received a $200,000 RAPID research grant from the National Science Foundation for its continued development.

Data from the wearable that is helping to keep the NBA bubble in play is also allowing researchers to use AI to better understand the spread of COVID-19. Scientists from WVU Rockefeller Neuroscience Institute are partnering with Oura, the innovator of a smart ring that monitors personal health data, to develop an AI model that identifies individuals with COVID-19 before they’re contagious.

4. Measuring the Effectiveness of Quarantine

What started out as a class project became something much bigger for one MIT student, who, alongside his professor, developed a machine learning model to assess the effectiveness of global quarantine measures. Using an SEIR model, the team trained a neural network to determine the number of infected individuals under quarantine, finding that “immediate government intervention in implementing strong quarantine measures” was most effective in flattening the curve.

5. Accelerating Treatment

For the one in 5,000 drugs that makes it to market, the discovery and development process can take upwards of 10 years and $2.6 billion to complete — 10 years and $500 million for a vaccine, under normal circumstances.

AI has the potential to change this by improving the efficiency of the drug development process. Relevant to the COVID-19 pandemic, it may also accelerate the search for viable treatment options that may already be commercially available. London-based BenevolentAI, for example, identified the rheumatoid arthritis drug baricitinib as a potential treatment for COVID-19 patients admitted to the hospital prior to developing critical lung damage, noting the drug’s anti-viral and anti-inflammatory properties. The firm’s findings were published in The Lancet and The Lancet Infectious Diseases, and the drug has received positive preliminary results in investigator-led trials.

The National Institutes of Health (NIH) is also harnessing the power of AI to better understand and treat the unique characteristics of COVID-19. Its newly launched Medical Imaging and Data Resource Center brings together cross-sector experts in AI and medical imaging to “create new tools that physicians can use for early detection and personalized therapies for COVID-19 patients.”

Looking Forward

Speed, accuracy and efficiency — mainstays of AI — have proven essential in identifying, managing and mitigating the effects of COVID-19. However, many of the areas outlined here are not unique to the current pandemic but have rather been brought to the forefront during this time of crisis. That makes these applications of AI even more intriguing as a means of tackling some of healthcare’s greatest challenges, including, hopefully, being better equipped for a future pandemic outbreak or disaster similar in scale.

Previous Article
 Why Health Plans Should Invest in Rapid Member Outreach Amid COVID and Hurricane Season
Why Health Plans Should Invest in Rapid Member Outreach Amid COVID and Hurricane Season

Healthcare Business Today

See more
Overcoming Programmatic & Operational Barriers to Medicaid Cost Containment
Overcoming Programmatic & Operational Barriers to Medicaid Cost Containment