Enlisting analytics and AI to contain the next pandemic
Article by SAS director of global government practice Steve Bennett.
Much has been written about the coronavirus since it was first identified in China in January and much more will undoubtedly be written before the subsequently alarming spread abates and medical science comes up with an effective cure.
And while news of the steady increase in reported numbers of people infected by and dying from COVID-19, as it is now known, has been dire, the good news is that we are getting much better at predicting and tracking the spread of infectious diseases.
Three out of four infectious diseases originate in other species but their rapid spread in humans is facilitated by our ever-increasing mobility. International travel is now such that a disease that might once have stayed relatively contained can now spread across the world in mere weeks. We saw this with the Severe Acute Respiratory Syndrome (SARS) virus in 2003 and we see it again today.
But the difference this time is the vast amounts of data we now generate in our daily lives. The smartphone didn’t exist in 2003, there was no such thing as the real Internet of Things, and the carriage of social media was infantile by today’s standards.
Given today’s proliferation of data sources and advances in technology, we are now far better equipped to address disease outbreaks. Analytics and artificial intelligence – and machine learning in particular – can mine and manipulate these data sources to help contain outbreaks in four phases of disease events.
Prediction. The human population is growing quickly and as we spread to different habitats, we interact with new species, and in different ways. This increases the opportunity for animal-borne diseases – of which there may be as many as 800,000 yet to be named – to jump across to humans.
By integrating data about known viruses with animal population and migration patterns, plus human demographics and cultural behaviour, AI has the potential to predict where new diseases could emerge. This enables the timely imposition of quarantine measures, and alerts health authorities to make ready to respond to outbreaks and even prevent them.
Detection. When an animal-borne virus makes the jump to a human, and possibly becomes airborne contagious, it can spread like a bushfire. So fast reaction is vital. To illustrate – one passenger on a three-hour Air China flight is estimated to have spread SARS to 20 people as much as seven rows away from him. Small wonder that coronavirus so alarmingly spread amongst cruise ship passengers.
As Director of the National Bio-Surveillance Integration Centre at the U.S. Department of Homeland Security, I oversaw the development of pilot projects that used machine learning to mine social media data – plus near-real-time emergency medical service and ambulance data – for anomalies in flu symptom reports. AI did a better job of predicting an out-of-the-ordinary disease event than traditional disease reporting, and did it a whole week faster.
The sooner an outbreak is identified, the sooner protocols can be put in place to stop the spread and heal the sick.
Response. There are two ways AI can help shape the response to an emerging epidemic. First, by leveraging any and all data that might be possibly useful. At the U.S. Department of Homeland Security, in addition to the obvious travel data that we mined to predict new types of flu, we even included data from the Australasian Flyway which is the migratory route taken by birds flying between China and Alaska.
Also, deep learning – another AI technology =- can improve treatments and accelerate the developments of new ones. For example, chest X-rays of coronavirus patients can be used as input to create AI models that speed diagnosis. Creating new vaccines and antiviral medications is a long process and subject to much trial and error. But AI can examine data from other diseases to predict what types of remedies are likely to be most effective.
Last year, the first fully AI-developed vaccine was created in Australia, saving years of effort and millions of dollars.
Recovery. When a disease has been contained, machine learning can help policymakers plan to prevent similar future outbreaks by performing ‘What if?’ analyses to simulate the impact of policies and initiatives. Doing this establishes the basis for data-driven decision-making with a better likelihood of preventing or containing the next pandemic.