Researchers use artificial intelligence to make coronavirus predictions

The technology has so far identified three key factors that can predict the future severity of a COVID-19 case.

Researchers at NYU and Columbia have developed a machine learning technology that predicts the severity of coronavirus cases.

NYU Grossman School of Medicine and Columbia University led a team of American and Chinese scientists to utilize artificial intelligence (AI) techniques in a new prototype that will predict the severity of lasting effects from COVID-19.

The researchers set out to demonstrate the effectiveness of artificial intelligence techniques in predicting which patients infected with SARS-CoV-2, the virus that causes COVID-19, would later develop potentially fatal acute respiratory distress syndrome (ARDS). 

[RELATED: NYU hospital to coronavirus doctors, nurses: Don't tell media about mask shortage]

The small-scale study consisted of 53 consecutive patients, 33 men, and 20 women, infected with SARS-CoV-2 who went to Wenzhou Central Hospital or Wuhan Central Hospital for treatment in January 2020. The median age was  43 years old. Mild symptoms at onset included cough, fever, stomach upset, myalgia, wheezing, dyspnea, nasal congestion, and sore throat. 

The study used a type of artificial intelligence called predictive analytics. “Predictive analytics...learns from historical data to help predict future outcomes,” the researchers explained. 

“The technology uses machine learning algorithms that can extract insights and rules from experience (historical examples) in order to determine data attributes (features) with the most predictive power for making accurate predictions,” revealed the researchers. The features tested include physical symptoms, radiological findings, and comprehensive blood analysis data.

[REALTED: Another medical research prof with secret ties to China]

The prototype found that the most accurate predictor of severe ARDS was a combination of changes in three features. These features include, in this order, “a mildly elevated alanine aminotransferase (ALT) (a liver enzyme), the presence of myalgias (body aches), and an elevated hemoglobin (red blood cells)."

The researchers reported that they were able to predict the risk of ARDS with up to seventy to eighty percent accuracy. 

The team of investigators admits that they would like to conduct further testing in the near future to “further [validate] and [refine] this model for a wider clinical spectrum.” They would look to increase the median age of patients tested and include a more diverse array of “disease presentations.” 

The researchers’ intent was not to “supersede clinical reasoning,” but rather to use predictive analytics to supplement doctors’ decisions in their diagnoses of “sick” and “not sick.” It also gives attention to some clinical data that could be “underappreciated” by medical professionals. 

“I'm very proud to see NYU taking the lead on this,” Bobby Miller, junior at NYU and Campus Correspondent for Campus Reform told Campus Reform

“I think it’s fascinating that NYU was able to come up with something so advanced during a time of crisis, and it’s remarkable to see the community join efforts with an international school to make it happen,” Liz Alvarez, a junior at NYU, told Campus Reform

Follow the author of this article on Twitter: @redwave1776 and Instagram: @enna.marie3519