Jameel Clinic Fuels Health Breakthroughs with Power of AI

MIT’s Building 19 might be an unassuming spot for the start of a medical revolution, but this is the home of the MIT Abdul Latif Jameel Clinic for Machine Learning in Health (Jameel Clinic), which—thanks to a gift made during the Campaign for a Better World—is applying artificial intelligence and machine-learning technologies to address major challenges in public health.

Launched in 2018, the Jameel Clinic is the fourth major collaborative effort between MIT and Community Jameel, the social enterprise organization founded and chaired by alumnus Mohammed Abdul Latif Jameel ’78. It brings together, in the words of its program director, Ignacio Fuentes, the “rock stars” of disparate fields—busy researchers who wouldn’t normally work together—and helps forge them into a community that’s achieving transformative results.

Jameel Clinic’s machine-learning model evaluated 107 million new molecules for potential antibiotic properties in just a few days.

Jameel Clinic’s machine-learning model evaluated 107 million new molecules for potential antibiotic properties in just a few days.

Take antibiotics. Drugs designed to wipe out bacteria tend to work well at first. But with time, genetic variation and mutations lead to drug resistance, enabling superbugs to outmaneuver pharmaceutical arsenals. The last time a new class of antibiotics was introduced to the world was 30 years ago, so innovation is desperately needed. Traditional drug development is an expensive, time-consuming, and challenging process featuring a lot of trial and error. Jameel Clinic has taken a wholly different approach, one that employs artificial intelligence to speed discovery.

In work reported in 2020, for example, Jameel Clinic researchers created a machine-learning model that analyzed the structures of 2,500 compounds with known interactions (or noninteractions) with bacteria to evaluate 107 million new molecules for potential antibiotic properties. The work, which took a matter of months, would have required 14 years to complete by hand in the lab, says Jonathan Stokes, one of the lead researchers on the project, a former postdoc at MIT and the Broad Institute of MIT and Harvard and currently an assistant professor at McMaster University.

“Machine learning is kicking medicine to the next level.”

Ignacio Fuentes
Jameel Clinic Program Director

One molecule emerged as a powerful new antibiotic—a drug the team called halicin, which is effective against many of the world’s most dangerous pathogens, including some that are resistant to all our current antibiotics. The halicin molecule doesn’t look or work like other drugs, which underscores the power of this research tactic: “The solutions suggested by the model were delightfully unexpected,” says Fuentes. Although more work needs to be done before the drug can be brought to market, “this approach is turning the traditional model of drug discovery on its head,” says Jim Collins, Jameel Clinic life sciences faculty lead and the Termeer Professor of Medical Engineering and Science. This “gives us a formidable new tool to fight both the microbial and viral foes we know and the ones we’ve yet to meet (like the next Covid-19).”

The new antibiotic molecule halicin was named after the fictional AI system HAL 9000 from 2001: A Space Odyssey.

Halicin was named after the fictional AI system HAL 9000 from 2001: A Space Odyssey.

Tools to Aid Diagnosis

Back in 2014, when Regina Barzilay, Jameel Clinic AI faculty lead and the School of Engineering Distinguished Professor for AI and Health, found out that she had breast cancer, she joined the legions of women worldwide who have confronted that diagnosis. She ultimately beat the tumor, but wondered whether it could have been detected earlier. So Barzilay worked with a team at Jameel Clinic and Massachusetts General Hospital to train a deep-learning model on 90,000 mammograms. The model discovered patterns within the images of breast tissue that successfully identified 82% more future cancers than the current standard of care.

Barzilay ran her own mammograms through the algorithm and determined the technology would have revealed her diagnosis two years earlier. Such work is significant for public health since early detection is crucial for treating and surviving breast cancer. “Rather than taking a one-size-fits-all approach, we can personalize screening around a woman’s risk of developing cancer,” says Barzilay. And she notes that the AI model also has the benefit of working equally well among different racial populations, a welcome finding given that Black women die from breast cancer at substantially higher rates than white women.

Jameel Clinic researchers see enormous potential for AI to improve health care across the globe. “This is a place where we can really impact people’s lives,” says Fuentes. “In just a few years, Jameel Clinic has already begun to transform the landscape of health care through the use of artificial intelligence. It’s clear to me that more breakthroughs lie just around the corner.”

Late last year, Jameel Clinic announced the ambitious AI Cures initiative intended to bring machine-learning solutions to find new antiviral molecules to combat Covid-19 and other emerging pathogens. In addition, it recently kicked off a three-year partnership with the Wellcome Trust, a charitable foundation, to expand the deployment of AI technologies globally, especially in developing countries with underserved communities. Jameel Clinic also intends to expand its predictive diagnostics to lung, prostate, pancreatic, and liver cancers; use AI approaches to improve outcomes for cerebrovascular and cardiovascular diseases; and encourage adoption of these technologies by regulators and health care practitioners alike.

“Machine learning is kicking medicine to the next level,” Fuentes says.

—Ari Daniel PhD ’08

Share this story