Can AI Save Us From Superbugs?

You've probably heard of artificial intelligence writing college admissions essays or producing strikingly authentic-looking "deep fake" images. Less discussed is AI's potential to help us address challenging health crises.

Indeed, if deployed properly, AI could equip scientists with tools to fight the rise of drug-resistant "superbugs."

Scientists predict that superbugs -- bacteria and fungi that have developed resistance to existing medicines -- could kill 10 million people per year by 2050. That would make drug resistance an even bigger killer than cancer.

Fortunately, AI has the potential to usher in a new era of rapid treatment discovery to give humans a leg up on superbugs. But it'll take the best efforts of the public and private sectors to ensure these new drugs are developed, approved, and accessible to patients -- before it's too late.

In 2019, superbugs were linked to the deaths of almost 173,000 Americans -- and nearly 5 million people worldwide.  

Despite this alarming trend, many large companies have stopped researching and developing novel antimicrobials. That's not due to lack of scientific promise. It's because the process is long, expensive, and often commercially infeasible due to the unique challenges of antimicrobial use. In fact, all of the small companies that received FDA approval for a new antibiotic since 2017 have either filed for bankruptcy, been bought out by another company, or shut their doors.

We need to attack this problem from two ends: optimizing the discovery and development of novel treatments and reshaping the broken antibiotic market with new incentives.

My company and our academic partners are working on the first part of the problem. Using AI algorithms, we're developing new classes of antibiotics that treat the world's most urgent threats. In days or weeks, AI can do essential discovery work that would take researchers months or even years.

Here's how it works. Researchers expose a pathogen to thousands of chemicals with diverse structures to determine which ones prevent bacterial growth. They use the results to train an AI model to predict which new chemical compounds might be similarly effective.

Researchers then bombard the trained model with millions to billions of possible molecular structures, homing in on ones that look most promising. AI can virtually screen millions of molecules in an afternoon, no petri dishes required, compared to thousands using traditional methods.

Scientists then test the most likely prospects, pairing biological with computational expertise. AI could shorten the time between drug discovery and the pre-investigation stage from roughly 4.5 to 2.5 years. Even considering the cost of the requisite computing power, AI could reduce research expenses to one-third of what they might be otherwise -- making it affordable to pursue antimicrobial treatments that may not have been possible with traditional methods.

With breakthroughs like these, we are poised to discover powerful new antimicrobials. Yet the unique economics of these superbug-killers mean there's little incentive to develop them.

Clinicians must use antibiotics judiciously to preserve their effectiveness. This, along with the short courses of therapy and low reimbursements for antibiotics, has contributed to challenging economics for companies to recoup the significant investments made to bring a new antibiotic to market.

Government efforts also have a key role to play. For example, the bipartisan PASTEUR Act was reintroduced in April of last year, and would create a subscription-like model to ensure that if a company develops a successful new antimicrobial, it will receive a sufficient return on investment while also supporting the treatment's appropriate use.

My colleagues and I are confident we can outpace antimicrobial resistance scientifically. But we can't do so on our own. We need a multi-pronged effort that includes cutting-edge approaches to early development and a reinvigorated marketplace. The fate of modern medicine depends on it.

Dr. Akhila Kosaraju is CEO and president of Phare Bio.