How AI is helping the fight against COVID-19

Experts from around the world discuss the benefits artificial intelligence and machine learning techniques are bringing to the battle to beat the pandemic, and the ways in which we should help the technology be more effective

Artificial intelligence has been used to good effect across the world to help scientists in their mission to beat the Coronavirus pandemic.

Speaking at a virtual roundtable organised by the International Telecommunication Union (ITU), Soo Jun Park, Assistant Vice President, ETRI (Electronics and Telecommunications Research Institute, Korea), explained how the technology was employed in his country, utilising lessons learned during the MERS (Middle East Respiratory Syndrome) outbreak of 2015.

"We were the worst-hit region besides the Middle East. At the time we didn't know how to cope with infectious diseases. One key problem we realise now was that at that time the government concealed all the information. It didn't release data about patients, hospital capacity, or anything. That exponentially increased the number of infected patients."

He continued, adding that the key lesson was about making the data widely available.

"We learned that to cope with infectious disease, openness is key, it's not just about gathering the data. You have to let people know what's happening, and let them know the truth about what's going on. We changed laws to accommodate that.

"Korea had our first confirmed case [of COVID-19] in January 2020, a lady who'd arrived from Wuhan. From that moment the government reacted very quickly, based on our previous experience. We also had the right IT infrastructure in place. We have very sophisticated medical systems, with universal health insurance for everyone. Those things combined helped us handle the pandemic."

That infrastructure includes a smart quaratine system which gatherings information on arrivals from countries where infectious diseases have occurred, and tracking and monitoring those diseases during the quaratine period. It also shares that data with medical institutions.

Korea also has a self-health check app for new arrivals, contact tracing, and an AI-driven COVID-19 X-RAY and CT image screening solution.

Moez Draief, global chief scientist and VP of data science and engineering at Capgemini added that AI is also used in background research.

"We're also using AI in the background, where it sifts through thousands of medical papers for relevant information, and helps our understanding of any potential adverse affects of any drugs we're developing."

He added that collaboration with regulators is also important, something which the UK could perhaps learn from.

"In Europe the regulators are very open to engagement to help innovation progress, there's an understanding that we have to work together. Scientists need to know where the regulator is happy for innovation to happen. And the regulator needs to anticipate where the innovation is likely to happen. We need this to avoid issues occurring down the line where either side is surprised. For instance contact tracing in France is done hand-in-hand with the regulator to understand how it should be done."

He also discussed how AI can help with any anti-vaccination sentiment.

"We are engaging to understand the arguments and which parts of the population are sceptical about vaccines and why. There's a lot of fake news out there. We have a project where we're looking at communities who have taken cancer drugs, and enabling them to interact, and to have tools to recommend people for them to talk to, to create a dialogue rather than an echo chamber. This is where AI can help, not to solve the problem but at least to facilitate the dialogue."

Ulla Jasper, Policy Lead from the Botnar Foundation, a philanthropic foundation based in Switzerland, said that more work needs to be done to enable data sharing on a global scale.

"There are lots of initiatives and voluntary codes, but too little concerted effort. Not that we need to arrive at a global one-size-fits-all data governance framework, but instead how can we share data across borders? The flow of data during Covid has not been perfect. We're not necessarily looking for one big solution, but we need to find sub-fields where we can see progress more easily. Like data sharing in public health emergencies."

Fred Werner, head of public engagement at the ITU sounded a similar note, calling for greater data sharing.

"There's more data now than ever before. But that doesn't lead to a greater exchange of data and collaboration. I've been in lots of different meetings where data is always the crux of the issue. Someone will ask who has data, and everyone raises their hands. Then someone asks who's willing to share it? And everyone looks at their shoes.

"So working out how to share it safely and securely is critical. There's not a lack of open source data, but you look at the basics like; is it discoverable, is it labelled correctly, is it freely available or licensed, or peer to peer? These are all bottlenecks."

As long as these issues are identified and acknowledged, we have at least taken the first step towards solving them. However, with COVID-19 still reaching ever higher peaks in many countries, the time to accelerate the next steps is now.