AI Drug Design Could Save Lives

You can hardly open a business publication without encountering rosy projections of how artificial intelligence (AI) will revolutionize industry. Pharma is no exception.

Developing new molecules takes years of effort and millions of dollars. What if we could accelerate this process by using AI to sift through thousands of studies, perform hundreds of experiments, and generate dozens of insights?

All in days, not months? Wouldn’t this lead to a wonderful world of effective — and cheaper — drugs?

Effective? Yes. Cheaper? Probably not. Let’s be real. New drugs are expensive because they have to be. Pharma is more than science, it’s also a business. Not only do manufacturers have to discover and develop new drugs — a great many of which will fail in development — they also have to reward the investors whose commitment makes the entire enterprise possible.

It’s capitalism 101, a fact many of our critics fail to grasp. A July 14 article in The New York Times moaned that pharma R&D expenditures are typically 10% lower than returns to investors.

What’s wrong with that? If rewarding stockholders offends our critics, perhaps they should divest publicly held companies from their portfolios and stuff the proceeds into mattresses. Face it, even as new drugs are discovered, they still need to be tested, manufactured, and marketed. And they must command prices that will keep companies solvent.

So if AI won’t cut the cost of drugs, what can it do?

Anybody remember the 2014 Ebola panic? I bet you know someone who was ready to fill a suitcase with Cipro and head for a cabin in the Catskills. Hard to blame them, though; plagues are scary.

The 1918 worldwide flu pandemic killed 50 million people, more than the combined populations of New York, Pennsylvania, and New Jersey. Even with today’s vaccines, 36,000 Americans die each year from the flu, and new strains appear constantly. Who knows what’s around the corner.

Unfortunately, our track record in responding to epidemics isn’t so hot. HIV vaccine? Still working on it. Zika? On the drawing board. Ebola? Any day now.

Instead of preparing to stop the next epidemic, our primary reaction today to infectious disease is shutting our borders to keep out those who might be infected.

That’s where AI can make a difference. AI’s ability to process seemingly infinite amounts of information including social media and local reports could permit earlier detection of new pathogens or dangerous mutations in existing organisms.

It took years for the potential impact of Zika to be recognized during which time it gained a foothold in much of Latin America. Could better surveillance have given us the jump needed to avoid thousands of cases of microcephaly?

AI complements what we in pharma have always done best — respond to medical needs quickly and find solutions. However, to embrace AI fully, we need to dismiss the myth that it’s dangerous. Ignore sci-fi images of malevolent computers taking over the world. AI is a tool that can help us master global health challenges in ways never possible before.

Think of it this way. The Earth has been socked many times in the past by meteors, and it’s highly likely to happen again. Until now, there was little we could do to meet this threat. Finally we’re acting — with AI programs that tirelessly scan the skies to detect potential “incoming” asteroids and comets. The opportunity to use AI for disease surveillance gives us the chance to meet another ancient threat.

Let’s turn the tables on new or expected diseases. Agree?

Sander Flaum is a principal at Flaum Navigators.