AI software from University of Waterloo researchers could help identify COVID-19 patients who need ICU care

Saskatchewan ICUs are being pushed to the brink as COVID-19 hospitalizations hit record highs.

WATERLOO -

Artificial intelligence software developed by researchers at the University of Waterloo (UW) could help doctors make decisions about whether COVID-19 patients need treatment in the intensive care unit.

The software, which was researched by UW staff and alumni-founded start up DarwinAI, looks at more than 200 clinical data points, including vital signs, blood test results and medical history of the patient.

“That is a very important step in the clinical decision support process for triaging patients and developing treatment plans,” said Alexander Wong, one of the researchers on the project, in a news release.

The software was trained using data from around 400 cases at Hospital Sirio-Libanes in Sao Paulo, Brazil.

According to the release, the neural network can predict ICU admissions in new COVID-19 cases with more than 95 per cent accuracy, along with sharing the key factors in its predictions.

Researchers said the tool is meant to help doctors make decisions more quickly, rather than replacing the need for a physician.

“The goal is to help clinicians make faster, more consistent decisions based on past patient cases and outcomes,” Wong said. “It’s all about augmenting their expertise to optimize the use of medical resources and individualize patient care.”

The technology is available for free so engineers and scientists around the world can work with it and improve it. It’s also being incorporated into the COVID-Net open-source initiative, which is helping doctors detect and determine the severity of COVID-19 through AI analysis of medical images.

A paper on the research will be presented at the 2021 Conference on Neural Information Processing Systems on Dec. 10.