The COVID-19 pandemic continues to have a devastating impact on the health and well-being of the global population. Against this background, Alexander Wong, a co-founder of DarwinAI, a startup in Waterloo, Ontario, Canada, and a professor at the University of Waterloo, together with his research student Linda Wang, developed an AI tool " COVID- Net ", hope to help guide medical professionals about the severity of COVID-19 infection.
COVID-Net is a deep convolutional neural network designed to screen patients with suspected coronavirus infection by identifying obvious signs of the disease on chest X-rays. As stated by ZDNet , although COVID-19 may be a relatively new infection, AI has long been used to analyze images of various medical conditions, including cancer and eye diseases.
In late March, COVID-Net has completed open source. Wong and Wang wrote in their COVID-Net paper, "It is hoped that the open-access COVID-Net and the description about building the open-source COVIDx dataset will be used by researchers and others, and can be used as a basis to accelerate development The speed of highly accurate and practical deep learning solutions to discover cases of COVID-19 while accelerating the treatment of those most in need. "
DarwinAI's CEO Sheldon Fernandez said that after COVID-Net's open source, "the response was overwhelming." They have received suggestions for improvement from many people, and these people have also explored and developed on the existing basis.
Since its release, researchers from Yemen, Indonesia, Turkey and Saudi Arabia have begun to experiment with COVID-Net. For areas where computing resources may be scarce (such as developing countries or poorly connected rural environments), DarwinAI has also released a "lite" version of COVID-Net, which can be run on a laptop. According to Fernandez, the accuracy of this version has only dropped a bit compared to the full version.
It is worth mentioning that in addition to DarwinAI, many companies including Alibaba in China are also conducting parallel research, hoping to use AI to identify COVID-19 infection on X-rays. However, the products of these companies are not yet open source.
Fernandez revealed that COVID-Net has not yet been put into production, but the company hopes it will be ready in a very short time. At present, this is just a very technical implementation that data scientists can use, and cannot be used by radiologists or medical staff. Therefore, the tool needs to be packaged in an appropriate application UI that is easy to use, reducing the technical requirements so that it can be used by more people. The company is working with many government organizations in Canada to try to obtain some funds to solve this problem.
He also said that medical institutions can use COVID-Net to classify patients, "we think it can be used as a free tool with more conclusive tests."
Currently, the COVID-Net team is working on COVID-RiskNet, a deep neural network tailored for COVID-19 risk stratification. The purpose of COVID-RiskNet is to predict whether the disease will seriously affect a person and what treatment they may need. "This is a tool that allows them to quickly understand the severity of the problem and then classify it accordingly." DarwinAI hopes to open source COVID-RiskNet in the coming weeks.