Artificial Intelligence and Drug Discovery—We Aren’t in Kansas Anymore, Toto

Author(s)Troy Trygstad, PharmD, PhD, MBA, Pharmacy Times Editor in Chief

Listen

0:00 / 0:00

Key Takeaways

Traditional drug discovery involves theoretical matching, chemical synthesis, and extensive testing, culminating in human trials over decades.
AI could revolutionize drug discovery by compressing timelines, reducing costs, and generating numerous drug candidates rapidly.
AI's computing power, enabled by advanced chips, could transform clinical trials and personalized medicine, raising regulatory and IP issues.
Quantum computing promises to redefine data security and computational capabilities, impacting AI and healthcare innovation.

The speed and scale of what’s coming with AI can only be imagined now, but without a doubt, our lives will change.

First There Was the Plow: Drug Discovery as It Exists Now

It Starts With Theory

For more than a century, drug discovery started at a chalkboard, with subject matter experts on a particular set of biological receptors and an attempt to match against a known or generally well-understood pathological pathway that would theoretically lead to some other set of biological responses to slow, stop, or ameliorate disease expressions and symptoms. Once a theoretical subject drug and human receptor or target tissue match was made, the findings were sent along to the chemists and bioengineers.

Then Creation of a Subject Moiety or Protein

For small molecules, call the organic chemists to concoct a manufacturing process that produces the subject chemical. For large proteins, call the bioengineers and have them figure out how Mother Nature can produce the proteins through eggs, field corn, or some other biological production process.

About the Author

Troy Trygstad, PharmD, PhD, MBA, is the executive director of CPESN USA, a clinically integrated network of more than 3500 participating pharmacies. He received his PharmD and MBA degrees from Drake University and a PhD in pharmaceutical outcomes and policy from the University of North Carolina. He has recently served on the board of directors for the Pharmacy Quality Alliance and the American Pharmacists Association Foundation. He also proudly practiced in community pharmacies across the state of North Carolina for 17 years.

Then to the Petri Dish, Tissue Lab, or Testing in Animal Models

Now the application of theory goes through real-world trials with biological systems. At this point, thousands of candidates could be whittled to tens or single digits of subjects that must feel like drilling for oil and gas to the investors. If one strikes, it is off to human subjects.

Then, Finally, Out to Safety and Efficacy Trials in Humans

Often tens or even hundreds of millions of dollars and decades later, into the human subjects the test moieties and proteins go for phase 1, 2, and 3 trials. Then, if all goes well, it’s off to the market, where we continue to monitor with phase 4 or pharmacovigilance trials with adverse effect reporting, and sometimes discovery of new uses (such as weight loss for glucagon-like peptide-1 agonists after a focus on diabetes, or new potential uses such as treatment of substance use disorder).

Then There Was the Tractor: Drug Discovery as It Will Exist in the Future

What if artificial intelligence (AI) could compress the expense and timeline of the first 3 steps before testing in humans? What if it could eliminate the need for them? Could AI compress 30 years of drug discovery into 3 days and spit out the next 100 blockbuster drugs after scanning all that humans know—every article, textbook, data point, or model run against every blood test, radiology result, adverse effect report, or social media post? If so, the bottlenecks that remain would be the speed with which a company could initiate and complete clinical trials, the regulatory process to go with it, and the tomes of patent submissions to the US Patent and Trademark Office that would likely follow in rapid succession. It is no wonder pharmaceutical manufacturers are looking to pharmacies for solutions to expedite clinical trial enrollment and deployment in a decentralized, rapid, cost-effective manner.

The Awesome Computing Power of AI, Leading to Nonbiological Experts of Everything, Everywhere, All at Once

Human Beings Struggle With Scale

Try this mental exercise: Think of a trillion dollars. It is big, right? But how big? Well, it is 1 million millions.

Was it easier to understand? Maybe telling voters that the US has $34 million million in debt might wake more of us up to the staggering amount of money associated with our gross domestic product, our economy, and our debts. Ever watched Men in Black? Remember that universe on Orion’s belt? The entire universe was hanging in a trinket on a cat…and of course, the fate of the universe was hanging in the balance, as well. Or at least the fate of 1 universe, that is. Apparently, that sort of stuff has been proven or theoretically possible.

Astrophysicists Are Smart People

As I age closer to 50 years, I find myself following physicist Brian Cox and his contemporaries (such as Neil deGrasse Tyson, with whom you may be more familiar). I find their layperson explanations of the laws of nature captivating as I look back on my life and its relationship with the universe. I certainly am in the camp of comfort with both science and deity being unconflicted.

The physics and astronomy books I grew up with have been completely rewritten. Most everything you see in a Marvel film, or the movie Interstellar, or The Jetsons, is either observable and now tested or theorized by some of the smartest people ever to walk the earth. Above all, scale fascinates me. There are more than 100 billion stars in our Milky Way galaxy and 200 to 2 trillion observable galaxies in a potentially infinite universe. Oh wait, and there may be multiple parallel universes, too. The element of nature that still gets me is looking out at the stars and realizing that what you see (to our eyes) is what those stars looked like billions of years ago, because the speed of light is so relatively slow compared with the task of reaching those distant stars, even traveling 186,000 miles per second across the universe (the speed of light). The scale of the universe is truly spectacular to us human beings. Then there is the theory of multiple dimensions and that dimensions may be contained within structures the size of a golf ball, much like the mass of a collapsing star. Scale is everything. If you understand scale and can harness it, you become a demigod.

AI is now possible because of computing power. The antecedents of neural networks go back decades, but modern processing chips have enabled staggering computing power. Oak Ridge National Laboratory (yes, the folks who helped bring you the nuclear bomb as well as multiple peacetime advances in telecommunications and health care) just achieved a computing capacity of 11.7 quintillion computations per second.¹ That’s 18 zeros (6 more than 1 trillion) and that is why AI is being born now. The new scale of computing is now nearly beyond our comprehension.

AI in Pharmacy Will Raise Countless Issues We Have Not Yet Contemplated

Last month, President Donald J. Trump announced a $500 billion effort to accelerate AI development in the US under the name Stargate with Oracle, OpenAI, and Softbank. Their stated goal is to break new frontiers but also to keep AI dominance in the US. That project alone has been predicted to put a strain on our national power grid.² The scale of the computing and computations continues to grow rapidly.

But what of the financing and who gets the intellectual property and what regulations should surround it? This is even more complicated in medicine because we have chosen, uniquely in the world, to have a dual public good blended with marketplace approach to health care financing and delivery. The possibilities of drug discovery, surgical procedures, and nanobots directed by AI are not the things of science fiction anymore.

What if AI could not only generate thousands of candidates per day but even customize therapy? Have a newly diagnosed cancer? Send the biopsy to www.wefindcures.com and a semiconscious AI being will analyze it and return it to your local oncology clinic with an attached AI lab report showing the recipe specifically for your cancer, all while considering concomitant diseases, pharmacokinetics, liver enzyme activity, and capacity to withstand adverse effects by also accessing electronic medical record and Apple Watch data over the past decade.

We have not even scratched the surface of what is to come.

Quantum Computing Is Around the Corner

Are you a big fan of cryptocurrencies because of their anonymity and security, with lock-tight encryption? Enter quantum computing, which can likely break any kind of encryption. Quantum computing uses subatomic particles to store, retrieve, and compute information and tasks. That is yet another mind-bender. This technology is so powerful (and even more mind-bending) that it makes understanding the linear and 3-dimensional scale of the universe seem trivial. We are truly entering a whole new way of looking at and interacting with the world(s) around us.