• Home
• /
• Opinion

AI might have solutions to fight the novel coronavirus

Techniques developed in non-medical areas can be applied for discovering drugs for the current outbreak.

A man wearing a facemask lies dead on an abandoned street of Wuhan. Emergency medical personnel in white protective suit are hovering around. On a normal day, this street in the industrial Chinese city with 11 million people is bustling. Now it looks like an eerie scene right out of a sci-fi film.
Such fleeting images of the novel coronavirus will be forever etched in our minds. It is a human tragedy that none of us looking at the pictures can remotely fathom.
More than 1,110 fatalities later, the medical world is still grappling with the cause and treatment for novel coronavirus. Despite warlike efforts to contain the new virus, it has infected more than 45,200 people and counting. We are fortunate that China can martial extraordinary resources, yet there are human limits to fighting a highly contagious disease.
This is where technologies like artificial intelligence can step up. When we are faced with an impossible situation, it is always useful to break down a problem. Simply put, we need AI to address detection, containment, and treatment of novel coronavirus.
Time is a rare resource during a disease outbreak. Speed often becomes the difference between life and death. AI tools can rapidly decipher the behaviour of a virus from mountains of data. For instance, who are most likely to be at risk? Does age matter? Does location have a correlation?
Startup BlueDot's detection of novel coronavirus was famously ahead of the Chinese health authorities' announcement. Its AI algorithm used global news reports, and data from animal and plant disease networks and official announcements to detect the disease outbreak. Using this information and airline ticketing, it was able to predict novel coronavirus' flightpath, originating from Wuhan and moving to Bangkok, Seoul, Taipei, and Tokyo in its early days. Answers often lie in adjacent areas. Scientists are using anonymised smartphones data to identify probable locations at risk of being affected by the disease.
Data scientists are pinning a lot of hope on the social media chatter for early detection. It was information gleaned from Chinese social media that was first brought to the notice of World Health Organization (WHO). Yet the early detection of a new disease using social media is dubious because the data is sparse and often inaccurate. 
However, for containment, AI can be more than just experimental. Cross-infection is very real for this highly contagious disease. So, medical workers are extremely vulnerable. This makes robots pretty useful. Some hospitals in the Chinese province of Guangdong are using robots to dispense medicines and food. The AI machines collect infected linen and medical garbage and even disinfect themselves.
They not only restrict cross-infection but also reduce the workload. Each of them can perform the medicine dispensing job of three medical workers put together. Their ability to read maps and plan the most effective routes make them efficient. In the interest of transparency, they can film the entire process. Nothing can be more comforting than a human care giver. But in extreme situations, where risk to human life is high, the patients can be treated by robots under close human supervision.
When medical workers get infected, the sheer shortage of protective gear has been a key factor. If we run out of face masks and production starts to fall behind demand, we need to find alternate industries that come closest in terms of raw materials, manufacturing process, and location. These could be asked to step up to the demand. AI algorithms trained with the right dataset should be able to give us the information along with specific company names. Some are already stepping up like Diaper maker Daddy Baby that is overhauling its production lines to manufacture facemasks. 
Needless to say, treatment is the million-dollar question. Plenty of startups that are affiliated to pharma companies are claiming that AI has helped them discover medicines for the disease. There is a lot of 'could' and 'should' in their findings. None of them are conclusive as yet, quite understandably because of the time constraint, thus making it hard for us to distinguish real findings from PR stunts. But the process they lay out give us an idea about how AI will come to the rescue in such times of crisis.
Notably, AI techniques developed in non-medical areas can be applied for discovering drugs for novel coronavirus. Techniques used for creating highly credible deep fakes is being used to come up with new treatment. It might sound dodgy at first, but it is actually brilliant, and initial findings from AI startup Insilico is encouraging. Using GANs or generative adversarial networks, instead of fake videos it is discovering new molecules that can fight the disease.
In some ways, the terrible tragedy of novel coronavirus has changed healthcare forever. Hospitals will become far more receptive to using robots to be better prepared. Given the vast amounts of data to analyse, humans are at a disadvantage to quickly solve the mystery of a disease outbreak. So, AI can definitely pitch in. But AI will still be assisting doctors, nurses and epidemiologists who will ultimately vet the machine's findings.
Shalini Verma is CEO of PIVOT technologies