Computer software developed by researchers at the University of Waterloo identifies vulnerable patients by studying previous cases of patients with COVID-19 with known clinical outcomes

New models of artificial intelligence (AI) can help physicians prioritize care by predicting which patients with COVID-19 are most at risk for death or kidney injury during hospitalization.

Computer software, developed by researchers at the University of Waterloo, identifies vulnerable patients by studying previous cases of patients with COVID-19 with known clinical outcomes.

“There is huge potential for such predictive intelligence models because they can greatly assist clinicians in determining who needs care most and most urgently to increase survival and reduce serious injury,” Alexander Wong said. professor of systems design. Engineering in Waterloo.

The new models are part of an open source project called COVID-Net, which has created several other innovations since the beginning of the coronavirus pandemic.

New models of artificial intelligence work by analyzing clinical and biochemical markers such as serum ferritin levels, the use of therapeutic heparin, heart rate and blood pressure, and automatically detecting patterns that predict patient death or kidney injury. Developed using explanatory artificial intelligence technology, the models also explain what indicators they relied on for their predictions, a major requirement to give physicians the confidence to act on results.

Explainers AI is a set of tools and frameworks used to interpret predictions made by machine learning models.

“AI models that provide not only predictions but also substantiate predictions can greatly improve credibility and dissemination to support clinicians in their decision-making processes throughout the clinical workflow,” said Wong, director of vision and imaging (VIP). ). ) Laboratory in Waterloo.

The researchers collaborated on the models with Dr. Adrian Florea, an ambulance doctor at CIUSSS de l’Ouest-de-l’Île-de-Montréal, and plan to test them in a clinical setting to gain further insight and improve their accuracy.

As in previous studies in the COVID-Net project, they have also made their work and results available to researchers and scientists around the world.

“Hospitals are already very overwhelmed by the pandemic, especially due to the recent jumps due to Omicron and its subvariants and recombinants,” said Wong, Canada’s Department of Artificial Intelligence and Medical Imaging Research. “The availability of artificial intelligence models that will help health professionals effectively and efficiently identify who needs care can significantly reduce the burden as well as the cost of health care.”

Researchers expect that the models will be easily transferred to other diseases and conditions, and they are already studying their use outside of COVID-19.

Hossein Abalebi and Maya Pavlova, both students of the VIP Lab, contributed to the project along with PhD student Andrew Greenowski and Mohammad Javad Shafi, also a professor of systems design at Waterloo.

Source by [author_name]