How AI joins the fight against coronavirus
We’ve seen nothing like it for a century, but can the battle against coronavirus be fought and won by Artificial Intelligence? Darren Yates explains.
The first six months of 2020 haven’t exactly been the rosiest way to ring in a new decade. If it hasn’t been bushfires blackening huge chunks of the eastern Australian seaboard, its coronavirus sweeping across the globe. Not to mention our crazy new relationship with toilet paper. But as the pandemic spreads, artificial intelligence and its components of machine-learning and data-mining are playing vital roles in the fight. We look at some key players and how you can learn to analyse infection rate data yourself.
ANALYSING THE EARLY DATA
One of the earliest published reports on the new 2019 novel coronavirus (originally dubbed ‘2019-nCoV’ and now ‘COVID-19’) was written by Chinese medical scientists and published in late-January 2020 in The Lancet medical journal. You can read it online at www.thelancet.com/action/showPd f?pii=S0140-6736%2820%2930183-5 (PDF, free). The report covers just the initial 41 patients detected with the illness up until January 2, 2020. Yet even with this small sample, the report authors were able to identify characteristics of the illness using various data analysis methods from high-tech DNA bioanalysis through to simpler statistical techniques. For instance, the authors found that those patients over 65 years of age were more likely to require intensive care. They also identified symptoms and their level of commonality amongst patients including fever, cough and shortness of breath.
BIG TECH APPLIES ITS MUSCLE
However, the authors acknowledged there were still considerable gaps in their knowledge of this new disease and that further work was needed. DeepMind is an AI company purchased by Google in 2014 and in mid-March, it announced it had taken the last 60 years of DNA and protein research and combined it with ‘deep learning’ to arrive at a best estimate on protein structures making up the COVID-19 virus. By understanding the proteins that power a virus, researchers are said to be better placed to find a cure.
Machine-learning is used here to speed up that process.
The DeepMind news came in the same week as word spread that the U.S. government was enlisting Silicon Valley to bring its AI skills to bear against the virus. Reports are these skills would be deployed on soon-to-bereleased research about the virus by the government, but also against misinformation spreading online before it goes viral.
BASIC ANALYSIS YOU CAN DO
While most of us typically don’t have access to the sort of cloud computing horsepower Google has to play with, you can perform simple coronavirus data analysis on your PC. Global coronavirus infection rate data is publicly available and tutorials online
explain how to use data processing and visualisation to better understand the data.
Now this is not to trivialise in any way the seriousness of the outbreak, but rather to highlight the importance of data in real-life situations like this. Machine-learning, data mining and ‘AI’ are all hyped-up, must-have skills, but the point of machine-learning and having all of this data is to gain knowledge and understanding from it. One of the key skills now being sought by employers is the ability to communicate that knowledge in ways that stakeholders – customers, your boss or even heads of government – can understand and use to make informed decisions.
Among these key communication skills is data visualisation, which is