The study answers: Who to hospitalize, whom to send home?

Researchers have developed an artificial intelligence tool to reliably predict which of newly turned positive coronavirus patients would develop acute respiratory distress. (The Statesman)

In a surprise, the study also found that the common symptoms that usually determine future COVID severity were not reliable.

The work was led by NYU Grossman School of Medicine and the Courant Institute of Mathematical Sciences at New York University. The Wenzhou Central Hospital and Cangnan People’s Hospital, in Wenzhou, China, also took part in the study.

The results of the study were published online on March 30 in the journal Computers, Materials & Continua.

The AI study on coronavirus

The researchers collected key data from 53 patients in two Chinese hospitals who tested positive for coronavirus in January 2020. The data covered the patients’ demographic profile, pathological tests, and X-ray observations.

According to the study, the progression of the disease was fairly uniform across this population of 53 patients. It started with the milder symptoms of a cough, fever and an upset stomach.

However, they all developed severe respiratory symptoms within a week.

Computer modeling

“Our goal was to design and deploy a decision-support tool using AI capabilities – mostly predictive analytics – to flag future clinical coronavirus severity,” said co-author Anasse Bari, a clinical assistant professor in Computer Science at the Courant Institute of Mathematical Sciences at New York University.

Furthermore, the researchers were looking for answers to the question: is it possible to predict with any degree of reliability whether a newly infected coronavirus patient will ultimately go on to suffer from Acute Respiratory Distress Syndrome (ARDS) – that deadly influx of fluid into the lungs?

The AI-based computer modeling successfully predicted the risk of ARDS with up to 80% accuracy.

Coronavirus symptoms: Not the usual suspects

The study also revealed that the following observations commonly believed to signal the future onset of ARDS did not do so accurately:

• Fever
• Above normal immune responses
• Lung images resembling ground-glass opacity
• Age and gender

However, the study homed in on other factors that were more accurate in predicting the odds that the patient would contract ARDS:

• change in the level of a liver enzyme known as alanine aminotransferase (ALT)
• reported myalgia
• hemoglobin levels

“We hope that the tool, when fully developed, will be useful to physicians as they assess which moderately ill patients really need beds, and who can safely go home, with hospital resources stretched thin,” said Bari.

Related Story:  Artificial Intelligence: AI Tool For Diagnosing COVID-19 Made Publicly Available For Development