AI Talk: Drug discovery in the age of COVID-19

May 8, 2020 / By V. “Juggy” Jagannathan, PhD

This week, I am focusing on the topic of drug discovery. It’s clear that COVID-19, has unleashed the collective intellectual might of the whole world to find a cure. When it comes to drugs, there are basically two main types: those which are prophylactic, i.e. preventive like vaccines, or drugs that treat the condition to ameliorate the symptoms or cure it outright.


Bill Gates gives a good picture of how vaccines are developed in his blog. He has a nice educational YouTube video to boot! Quite interesting to watch floating viruses and antibodies that engage in a complex dance. He notes there are 115 different COVID-19 vaccine candidates and, if we get one by the beginning of next year, that would be a first in the history of vaccine development. Of note is the role genetics and AI in the vaccine development field. By starting with the genomic footprint of the virus, one company managed to come up with a vaccine design in 3 hours, as we noted in an earlier blog.


Over 500 clinical trials that have the potential to treat COVID-19 have been launched – this site tracks them all. In this frenetic search, a number of treatment suggestions have been derived with the help of AI. A company called Exscientia is combing through a library of 15,000 molecules with AI to see which one has the best chance of treating COVID-19. In another effort, AI was used to zero in on an arthritic drug as a good candidate for treating the virus. Which of these treatments eventually prove useful, only time will tell. In this effort, led by Lawrence Livermore National Lab, they use machine learning platform to design COVID-19 antibody sequences. Currently the only way to get antibodies is from people who have recovered from COVID-19. In this experiment, which consumed 200,000 CPU hours and 20,000 GPU hours, they started with an initial antibody candidate that was effective for an earlier version of coronavirus, and combed through the nearly infinite set of possibilities to arrive at 20 antibody candidates. What is a bit of a coincidence here is one of the supercomputers used for this experiment had the moniker Corona!


By delivery, I am referring to the delivery of medications to the home of patients! Last year we looked at how Amazon and other tech giants were exploring the use of drones for delivery. Now CVS, in partnership with UPS, has actually launched a service that will deliver medications using a drone. The initial target for this service is a large retirement community in Florida. Interestingly though, the last mile is done by a human carrier. This model has legs—literally and figuratively—and it can easily be replicated in all manner of places for the delivery of all manner of things—not just meds. I used to wonder how they would manage delivery using drones in big cities with large apartment complexes. By foot is the answer, at least for the last mile!

I am always looking for feedback and if you would like me to cover a story, please let me know. “See something, say something!” Leave me a comment below or ask a question on my blogger profile page.

V. “Juggy” Jagannathan, PhD, is Director of Research for 3M M*Modal and is an AI Evangelist with four decades of experience in AI and Computer Science research.