Computer Scientists Are Building Algorithms to Tackle COVID-19
Algorithms that can detect infections, differentiate COVID-19 from the common flu, and more
More than 125,000 people have been confirmed to have the novel coronavirus, COVID-19, around the world and the number is likely to drastically increase, according to health professionals.
Computer scientists and machine learning researchers are tackling the pandemic the way they know how: compiling datasets and building algorithms to learn from them.
There’s already a dataset of COVID-19 cases on Google’s data science competition platform Kaggle, which is updated with new cases daily. The data is robust, including patient age, location, when they started to experience symptoms, when they were exposed, when they entered a hospital, and many more. Nearly 300 people have used the data in their own analyses.
A researcher from the University of Montreal has collected and published a database of dozens of CT scans and chest X-ray images. The images are taken from publicly available studies on the disease.
And Johns Hopkins University has built an impressive dashboard of well-sourced data that’s updated regularly, giving a global look at the spread of the disease and its mortality. It can be copied and modified as the code is available on GitHub.
Other datasets have come directly from hospitals treating patients, which have quickly tried to turn around machine learning models to assist doctors looking for signs of the disease.
Here are some of those papers:
Shanghai researchers have devised a system that, alongside a human checking the results, could reduce the analysis time of a CT image from hours down to about four minutes.
Rapid AI Development Cycle for the Coronavirus (COVID-19) Pandemic: Initial Results for Automated Detection & Patient Monitoring using Deep Learning CT Image Analysis
This paper also claims to detect the presence of the COVID-19, but also visualizes the virus’s effects on the lungs to track the progress of the illness over time.
Abnormal respiratory patterns classifier may contribute to large-scale screening of people infected with COVID-19 in an accurate and unobtrusive manner
Researchers here look for an auditory way of screening for COVID-19 by analyzing how fast a person is breathing. The research isn’t conclusive, but it’s a new idea for a less invasive way of testing for the virus.
This work tries to differentiate the pneumonia suffered by patients with COVID-19 from the garden-variety flu.
Prediction of criticality in patients with severe Covid-19 infection using three clinical features: a machine learning-based prognostic model with clinical data in Wuhan
Using nearly 3,000 electronic health records from patients in Wuhan, China, researchers built an algorithm that could predict the rate of mortality for patients with more than 90% accuracy.