BIO: Exploring COVID-19's opportunities for AI, regulatory flexibility and 'handshake' deals

The challenges posed by the novel coronavirus are also providing new openings for rapid advancement in the life sciences field—especially in the adoption of big data and artificial intelligence, as companies look toward partnerships to tackle each scientific problem as fast and efficiently as possible, according to a panel of industry leaders at BIO’s virtual meeting.

“If you want to accelerate discovery, you look for the intersections,” said FDA Principal Deputy Commissioner Amy Abernethy.

“And I'm struck that COVID-19 gives us the opportunity right now to simultaneously develop our datasets that understand deep biology and the immune system, and our data sets that understand … what happens in the context of complex healthcare delivery systems that sometimes fail,” said Abernethy, who also serves as the agency’s acting chief information officer.

“What I've seen across the course of my career is some of the most interesting opportunities happen because we are pulling together streams of information, streams of talent, and capabilities that oftentimes aren't existing in the same spaces,” she said.

But effective research collaborations require all participants to have an equal seat at the table—physicians, data scientists and computer engineers alike—in order to reach those coveted eureka moments, Abernethy said.

At the same time, regulators have to keep pace with innovation, even outside the context of a global pandemic. And that can mean taking some chances.

“We have all had to figure this out really fast,” she said. “Companies are on development cycles where they are moving quickly and realizing that it's really important to talk to the FDA early and often—and that it's really important to take a few risks, especially if those risks are calculated and you have understood how to best manage them.”

“Meanwhile, on the FDA side, we've realized that we need to have regulatory flexibility in many different directions, and then learn from that regulatory flexibility on the fly,” Abernethy said. “We've issued a number of emergency use authorizations that taught us what is OK to do—and sometimes it doesn't work, and sometimes we have to pull back.”

That includes the agency’s policy on antibody blood tests, which had allowed companies to distribute diagnostic kits without federal review during the early stages of the disease’s spread. 

The FDA reversed that decision in early May and later listed more than two dozen tests to be pulled from the market, while granting nearly 20 antibody diagnostics to date full emergency use authorizations (EUAs).

“For example, what you saw with the serology EUA where we said, ‘You know, we actually probably now need to back off, and make sure we have better information about the performance of serologic tests before we maintain this particular emergency use approach,’” she said.

“So, as the FDA, we've been right-sizing right along with companies, and I think COVID-19 gives us this really important opportunity to try and ultimately figure out how do we innovate better—especially in spaces that are, historically, incredibly risk-averse.”

RELATED: FDA publishes first validation results of 12 COVID-19 antibody tests

Abernethy and the other panel members also discussed how an explosion in the amount of available cloud computing power has changed the biomedical research landscape in the past few years.

“During this pandemic response, we've been donating access to our computing infrastructure to lots of biotech companies,” said Peter Lee, Microsoft’s corporate vice president of AI and research. “And one biotech company, just doing some atomic-scale, thermodynamic modeling of the coronavirus’s spike protein—just that one job, and they're one of many—was consuming the same amount of computing power over a five-week period equivalent to 250 times the largest supercomputer during the first SARS epidemic.

“If we do manage to just compute our way out of this, it will feel lucky that it happened now and not five years ago because we just have that much more computing power,” Lee said. 

RELATED: Microsoft launches 5-year, $40M AI initiative in global health

One of the main drivers behind Microsoft’s growth as an AI services provider to the life sciences industry was an internal shift in sales policy, Lee said. Instead of representatives earning a commission based on the size of a particular contract, bonuses would be tied to how much those services were used. “That incentivized everyone to work with clients to make good use of it,” Lee said.

One of those largest contracts is with Novartis, launched last October, which has promised to put Microsoft’s AI tools on the desk of every research associate in the company over the next five years.

“This is not the first time a tech company has tried to partner with a pharma company—I think this is a level deeper,” said Shahram Ebadollahi, the drugmaker’s global head of data science and AI.

“This is not just selecting two projects and putting the best of the best of the two companies on them. This is collaborating with each other over several years, to really embed that kind of technology and have the mindset shift that we need within Novartis,” Ebadollahi said. 

“My real hope is that we can distill machine learning into bite-size engines, and infuse them in every place in the workflow—such that scientists that are coming up with the next molecule, the next manufacturing process, the next drug, and so on—that they don’t even realize that they’re using machine learning and AI,” he said.

Amgen’s senior VP of global research, Raymond Deshaies, echoed that sentiment, especially in the context of the enormous amounts of data being crunched in human genetics. 

“It came in initially through discovery research as a tool for discovering new targets—I believe we need to push it through the entire chain of medicine development,” he said. “But that’s going to be challenging because our medical system is so fragmented, and capturing that data is inherently difficult.”

RELATED: Microsoft joins Adaptive’s pursuit of blood-based diagnostics

Deshaies also talked about Amgen’s COVID-19 antibody research project with Adaptive Biotechnologies—whose CEO, Chad Robins, also sat on the panel—describing the pace of the deal as so fast they’re “building the airplane while they’re flying.”

“The little origin story is that [Amgen CEO Robert] Bradway and I had started talking about essentially using our screening technology and partnering with their antibody discovery manufacturing and commercialization during the Ebola crisis, but that didn't become a pandemic,” said Robins. 

“So as soon as we recognized what this was, we jumped into action basically on a handshake—and [Deshaies’] team has been helping us to develop our capabilities.”

First reported in early April, the two companies said they would “finalize financial details and terms in the coming weeks” in the rush to develop a neutralizing antibody treatment that could serve as a bridge to a potential COVID-19 vaccine.

And while the U.S. government is aiming to develop a vaccine by the end of the year, Abernethy says such accelerated timelines for research may not be all that far-fetched with some of the tools we have now.

“I reflect on the fact that, in 2000, I was caring for people with lung cancer by basically deciding small cell versus non-small cell, and that early-stage disease got surgery and radiation while late-stage got chemotherapy—that was about all we could do," she said.

“Twenty years later, we now have the ability to modulate the immune system, the ability to think about the genomics of the tumor and the genomics of the individual, the ability to think about social determinants of health, and how we put all that together to personalize treatments.

“What I'm excited about now is that it's not going to take 20 years anymore to solve that equation—it should be something less than a year, and we're going to try that out in the context of COVID-19. And I have incredible optimism that this is a reality that we're moving towards.”