The World Health Organization called it the world’s first “infodemic”. Social media has made it simple to spread fiction about the Coronavirus (or COVID-19). And when you stir up all the mud from the bottom of a pond, it’s difficult to get clean water. Especially in places where the spread of the virus is serious, people are having trouble determining what information to trust. This has led to panic where it’s unneeded and apathy where people should really care.
This is what happens when algorithms curate our news feeds, allowing anyone with an Internet connection to be heard.
Information aside, there are other roles AI has had in this pandemic. We already covered the company whose AI was able to identify the virus outbreak 9 days before the World Health Organization and 6 days before the CDC.
How, then, is AI being used to prevent the spread and even find a cure?
AI Drug Discovery
We’re amidst a time where computational drug design is possible. Algorithms can sift through massive troves of BioData to find existing molecules that might be able to fight COVID-19. Then, predictive models can be used to test their effectiveness. All of this makes it possible to discover and deploy drugs faster and cheaper than ever.
At least four AI-based proposals have been made towards treating, primarily inhibiting this coronavirus. Given the urgency of the situation, the initial drive has been toward finding existing, approved drugs that might be used.
Paul Dempsey, E&T
Inhibiting COVID-19 means preventing it from spreading within the body. Not necessarily curing the virus. An inhibitor would fill up all the molecular binding sites, preventing COVID-19 from binding to cells and reproducing. It’s kind of like being at a dinner party and all the seats are taken, so you cannot sit down and eat.
Some of the inhibitors discovered through machine learning processes include:
BenevolentAI, a UK-based start-up, has described baricitinib, which is used to treat rheumatoid arthritis, as one possibility. Deargen, a South Korean AI drug discovery specialist, has suggested atazanavir, used in HIV treatment, as a potentially promising option. Four more have been put forward by a team from China’s Army Medical University based on “high-throughput screening”.
Paul Dempsey, E&T
A fourth proposal, from Insilico Medicine of the US, has identified six new molecules that could be synthesised and tested to form the basis of a treatment.
What’s particularly fascinating is that this entire race to find the cure can be worked on by anyone thanks to many open-source tools:
- A Chinese Health Center published the fully-sequenced genome of COVID-19 on a public database called GenBank. Essentially allowing anyone to hypothesize and experiment.
- An open dataset on Kaggle is tracking the confirmed cases and locations of the COVID-19 outbreak.
- Modeling software such as AutoDock Vina allows developers to run molecular simulations.
What’s so exciting is that because many of these tools are open-source, in theory, some teen in Iowa could find a cure from a computer in their mom’s basement.
Siraj Raval, for example, is using his YouTube platform to launch a public initiative called the Coronavirus Deep Learning Competition. He’s doing his part to inspire his community of developers to work on this problem and he’s showing them numerous examples of how researchers are using GANs and Transformer models to discover possible COVID-19 inhibitors.
If you recall, GANs (or Generative Adversarial Networks) are the same type of algorithm used to create deepfakes. And Transformers the natural language processing algorithms I’ve used before to create my very own AI Writer.
That’s very innovative thinking in and of itself.
I believe that we’re looking at a paramount moment for machine learning in drug discovery – pushing it further along in its acceptance as a tool for improving how we discover new drugs.