SINGAPORE: There isn’t a conversation going around these days that doesn't touch on the subject of COVID-19 – and the virus spreading to dozens of countries around the world.
People are in fear, rightly so. People are worried for their own safety and their loved ones.
Healthcare workers and policymakers are figuring out the best ways to contain the virus while the public does what it can. Non-profit and grassroots organisations come together offering assistance.
Grab recently piloted a programme that provides round-the-clock rides to healthcare workers, a show of camaraderie on the part of drivers in this trying time for Singapore.
In my recent Grab rides, drivers are offering the use of hand sanitisers for free, even though one driver told me he paid a fortune for a bottle.
HOW ARTIFICIAL INTELLIGENCE CAN HELP WITH EARLY WARNING AND DETECTION
A larger question that some people have pondered is whether a technology solution could have detected and prevented a large-scale spread of a contagious disease like COVID-19. There is precedence.
A Canadian start-up called BlueDot used its proprietary artificial intelligence technology and natural language processing to canvas a vast amount of information to look for signs and predict where an infectious disease will turn up next. BlueDot scoured through 100,000 news reports in 65 languages on a daily basis.
The result: BlueDot sent an alert to its clients to avoid Wuhan on Dec 31, 2019, two weeks before the official announcement from the World Health Organization on Jan 9.
Using global airline ticketing data, it also predicted that the virus would spread to Seoul, Bangkok, Taipei and Tokyo primarily.
BlueDot is not without a track record, it had also successfully predicted the SARS pandemic.
This is certainly an interesting usage of artificial intelligence and machine learning beyond the standard tools employed by e-commerce and social media to throw up shopping recommendations or provide better search results capabilities. The question is how to have this information readily available to the public and relevant organisations.
There is a similar early warning tool Ushahidi developed in Kenya by a non-profit organization headquartered in Nairobi. Ushahidi uses crowdsourcing for social activism and public accountability, combining citizen journalism and geospatial information.
Ushahidi allows people to submit reports through SMS, apps, social media and the Internet, creating a temporal and geospatial archive of events. The Ushahidi platform is often used for crisis response, human rights reporting and election monitoring.
In Singapore, AI has also been enlisted to aid with detection. AI-powered temperature screening equipment piloted at Serangoon North and St Andrew’s Community Hospital in Simei does away with the need for manual screening, often time-consuming and manpower-intensive, and can detect and alert staff to individuals with high temperatures even if they were wearing spectacles or headgear.
RESEARCH INSTITUTES ARE ALSO PART OF THIS ECOSYSTEM
John Hopkins University is the leading institution leveraging the geographic information system (GIS) technology – this is a system designed to capture, store, manipulate, analyse, manage, and present all types of geographical data.
GIS technology uses data-mining to detect areas where people talk about the disease and creates heatmaps. These maps can help healthcare professionals and other key stakeholders in better tracking and zooming into a specific location to tackle the spread of a disease.
Harvard Medical School professor and Chief Innovation Officer at Boston Children’s Hospital John Brownstein had his team build Healthmap after the 2003 SARS epidemic, which scrapes news reports, chatrooms and more to build a visual picture of how the coronavirus is spreading.
It supplements data-gathering techniques by governments around the world and is also used in the WHO’s Epidemic Intelligence from Open Sources Initiative.
Another area technology could help us in is to make sense of the misinformation spreading online fuelling unnecessary hysteria and fear – and in some cases xenophobia – sometimes pushed by bots to create the false impression of many people talking about a particular subject and launch coordinated campaigns.
The use of a Botslayer prototype in the 2018 mid-term elections aided the Democrats in investigating Tweetstorms, their content and promoters to identify and ultimately take down malicious accounts.
Such tools can help journalists discern trending topics from surges that appear related to bot activity.
It’s during these trying times that technology should be a force for good that brings humanity together. AI algorithms can sort through which web pages tend to be accurate, which are salacious, and most importantly which posts likely come from bots rather than reputable sources.
READ: Commentary: What to do with all these health rumours and forwarded messages in the time of COVID-19?
BUT BEWARE TECHNOLOGY’S HUBRIS
There is one caveat however. All of this data-driven technology is premised on the validity and quality of the information it’s built upon. There is an old saying and it is still true in this case: Garbage in garbage out.
We should take heed from past cautionary tales. In 2008, researchers at Google claimed they could predict the flu trend based on people’s searches.
The idea was based on the assumption that people would search for flu-related information on Google when they caught the bug. “We can accurately estimate the current level of weekly influenza activity in each region of the United States with a reporting lag of about one day,” the Google scientists wrote.
But then this project failed, missing the peak of the 2013 flu season and failing to predict the 2009 H1N1 pandemic.
What this shows is — what Wired magazine calls “big data hubris” – garbage in leads to garbage out. Just because people searched for flu-like symptoms on Google doesn’t necessarily mean they have the flu. The vast majority of visits to the doctor for flu-like symptoms generally turned out to be other viruses.
In the same way, we can expect AI’s accuracy in detecting a virus to be less than desirable – but with machine learning, the hope is for each incident to provide constructive feedback to strengthen the algorithm in predicting future incidents.
A BRIDGE SINGAPORE CAN CROSS
Singapore with our strong push for AI and machine-learning development has an opportunity to harness and further support the development of such technological advancements that could determine how we anticipate and more effectively deal with the next health scare.
We have done a great job in coming together a society to fight a virus, but if we could leverage technology more pervasively to detect deceases early, understand it, and prevent mass hysteria – this would be a game-changer for us.
But this is admittedly easier said than done. Building up these types of AI, machine-learning, and data-mining capabilities require a huge and concerted financial investment, and potentially further research, development and test-bedding in order to produce a technology policymakers, businesses and clinicians can use.
With the drawdown of the Research, Innovation and Enterprise 2020 fund, no doubt businesses will look to Budget 2020 to see if further investments in this area can be co-funded or offset by government grants.
It is unfortunate that the world is facing yet another health scare – but we can rise above this just like we have in the past.
Singapore could take advantage of our existing knowledge and know-how to use this outbreak to test out prototypes and collaborate across nations to address the current COVID-19 and prevent another the next.
Jonathan Chang is a tech entrepreneur, investor, advisor, and lecturer.