Coronavirus outbreak the focus of artificial intelligence that is helping predict where it will strike next
People’s Google searches, social media posts and even chatbot questions are being used by artificial intelligence to try and predict where the novel coronavirus is going to pop up next.
The technology, which has been fine-tuned over the last 15 years, is already feeding information to major health agencies like the World Health Organisation to help them decide where they should focus their efforts.
One system, called HealthMap, uses publicly available data from across the internet as well as user-submitted information, according to one of its developers, John Brownstein, a professor at Harvard Medical School.
How AI is stalking coronavirus
“We work in this hybrid of data mining as well as crowdsourcing,” he told the ABC’s news podcast The Signal.
“What’s really phenomenal here is we’re seeing incredible international collaboration and a huge amount of data sharing.”
Search terms and social media posts feed the model
The technology Professor Brownstein and his team have created trawls everything from posts on medical websites to news and social media for signs of people talking about coronavirus.
If it finds people complaining of a high temperature or other symptoms associated with the strain of coronavirus — now named COVID-19 — before flagging the activity as more or less relevant to this outbreak.
“The system has done really well but we’re constantly retraining it, especially when you move into new languages and geographies,” Professor Brownstein said.
As technology permeates deeper into our lives, Professor Brownstein’s team are finding themselves with more data today than when they began the project 15 years ago.
“There’s more people interacting with symptom checker chatbots, so that’s a whole new source of data,” he said.
“There are people that are utilising connected devices, like a connected thermometer — that’s a very useful dataset that we’re getting access to.
“Voice technologies become one that we spent a lot of time on, working with Amazon, as an example, and the Alexa platform.”
While access to internet-connected thermometers and smart speakers might spark privacy concerns, Professor Brownstein said his team were careful with data they used.
“We have to think about patient privacy and anonymising datasets, that’s a big part of what we do,” he said.
“The good thing is, we work with a lot of public datasets, like open social media sites like Twitter which are already quite open and then we’re also doing a lot of work in crowdsourcing.
“[Crowdsourcing] is in our view the holy grail because we’re building platforms where people can engage, put their symptoms in, but be directly involved in public health.”
The artificial intelligence born from SARS
Professor Brownstein’s HealthMap was born in the shadows of the SARS outbreak in 2003, which killed almost 800 people and sparked panic across the globe.
At the time, Google Maps and social media platforms were just taking shape and his team believed there was an opportunity to collate the information into a single model.
“This idea of a mash-up that organised all the world’s information about disease events,” he said.
When the teams switched on HealthMap, it started to find virus and disease outbreaks all over the world.
“[We] trained it in a way to produce a living, breathing, real-time map of disease events that was sort of never seen before,” Professor Brownstein said.
“At the time, this seemed to be sort of almost looked down upon as a source of information that wouldn’t be as reliable as what government agencies were putting out.”
Soon the team realised the map was more useful than that and that it could actually be a tool for policymakers.
“Something flipped and that data actually became a huge driver of what government agencies used to understand what’s happening both within countries and in neighbouring countries and internationally,” Professor Brownstein said.
He said the real value of the technology, at least in the early stages, came from being able to identify outbreaks in countries that did not have a record of open transparency.
“In some ways [this AI could] apply this newfound pressure for government agencies to be more transparent because the information was going to come out one way or the other.”
More technology tracking the outbreak
As well as HealthMap, there are also other tools using algorithms to crunch huge amounts of data in an attempt to track coronavirus.
Ricardo Soares Magales, an associate professor in population health and biosecurity at the University of Queensland, told The Signal that digital epidemiologists were able to work out where the virus could pop up by combing everything from mobile phone locations to airport departures.
“These models are being updated pretty much every day as information comes in,” he said.
“Some of these models integrate movement of individuals between cities and between countries so that you have an understanding of how far the infection can go.”
“If we know that a particular location has a number of cases popping up and it’s highly connected to another location, we can inform services to target the surveillance and case ascertainment, the containment and isolation, in the receiving location.”
As well as tracking the spread, researchers are also trying to identify the source of the outbreak.
One theory has been that bat products sold from a Wuhan market were the origin of this outbreak, and Dr Magales said researchers were trying to get access to apps that people could have used to sell and trade wild animal products.
“Using that information to understand where the sale has been happening and who has been purchasing [the products] — not from an individual point of view, but an aggregate, [to see] which locations seem to be more active in the purchase of these products,” he said.
“Now authorities from the animal side are sieving through datastreams which could point to where these bat products have been and if they’ve been purchased online, which locations have been more exposed to this material.”
Professor Brownstein has been surprised by the number of people who have made the fight against the coronavirus a priority over other work.
“I think people have put their academic aspirations on hold while they’re actually trying to figure out and work together to get this thing right,” he said.
“[It] is very unique compared to other events we’ve seen.”
This content was originally published here.