Mayo Clinic Platform, the data analytics and digital health arm of the Rochester, Minn.-based medical center, is trying to stop diseases in patients before they emerge, its leader told Becker's.
"Do you remember the film Minority Report?" asked John Halamka, MD, president of Mayo Clinic Platform. "The notion in Minority Report was pre-crime — predicting the crime before it happens. What if we could predict disease before it happens? Then we could make sure that a patient receives appropriate interventions, treatment and monitoring so they don't experience severe disease."
One way the platform is doing that is through an algorithm, given FDA breakthrough status in May, that uses Apple Watch data to detect pulmonary hypertension in people who may not be aware they have it. The team at Mayo Clinic is also researching the use of machine learning for specialties such as radiation oncology, behavioral health and gastroenterology, among others.
The goal of the platform at its 2020 launch was to turn "healthcare data into wisdom," Dr. Halamka said. "And that is going to result in less invasive care, more timely care, less costly care and, ultimately, better wellness."
'First in healthcare to do that'
In early 2020, the platform deidentified the data of 10 million patients using the work of experts from Vanderbilt University and the Massachusetts Institute of Technology. Mayo Clinic has birth-to-death records of its patients going back more than a century.
"Mayo has one of the largest depths and breadths of deidentified, multimodal data for machine learning, but a set of processes that ensure that data never leaves Mayo Clinic and is used to innovate but protects and respects patient privacy preferences," Dr. Halamka said. "And that is pretty unique. I would say we're probably the first in healthcare to do that."
The platform contracts with digital health companies for the use of its deidentified data. One recent example is K Health, which has an AI algorithm to help triage hypertension patients. The firm had patient data from Israel but now uses Mayo Clinic's data to get a more accurate picture of its U.S. client population.
Personal experience with artificial intelligence
Dr. Halamka said his wife, now 60, was diagnosed with stage 3 breast cancer in 2011.
"I asked her: 'What if at the age of 45, you were told that an algorithm with 84 input variables could run on your phenotype, genotype and exposome and tell you with a pretty good degree of certainty that you would develop breast cancer in the future — would you take Tamoxifen today?
"(The medication) would cause a chemical menopause. And with menopause, there are certainly issues of muscle cramps and pain. There are side effects, right?
"And she said, 'You know what, having gone through cancer care, I actually would.'"
Dr. Halamka said he has been screened by Mayo's electrocardiogram lab, which can predict the percentage chances that patients will develop a variety of heart conditions.
"Basically, these algorithms were able to provide interpretations about future disease that were quite reassuring," he said. "My supraventricular tachycardia is something that is actually pretty benign."
He also has glaucoma, which runs in his family.
"Mayo was quite concerned that I was losing vision in my left eye a little faster than I should," he said. "So they did an MRI of my brain to figure out if there was a tumor or some ventricular problem, something unusual. Mayo has loaded 220 million images into Google Cloud and created the first-of-its-kind range of normals — evaluation of what we call body composition: fat and muscle tissue — understanding what a normal X-ray really is.
"And it was that work done in Google Cloud by Mayo that assessed that my brain was actually within the acceptable range of normal."
Ethics, innovation make up future work
To address ethical concerns about this futuristic approach to healthcare, Mayo Clinic formed the Coalition for Health AI with organizations including Stanford Medicine, Johns Hopkins University, the University of California, Duke University, HHS, the FDA, Google and Microsoft.
"I could reassure people by saying: Machine learning is math, not magic," Dr. Halamka said. "It is purely offering just another data point to a clinician that helps them understand the patient in front of them based on the data of patients in the past."
Before joining Mayo Clinic, Dr. Halamka, a practicing emergency medicine physician, served as CIO of Beth Israel Deaconess Medical Center in Boston. He was also a healthcare innovation professor at Harvard Medical School, where he advised the Obama and George W. Bush administrations on health IT strategy.
Dr. Halamka taught himself how to use analog, digital and early microprocessors before he was a teenager. He sold his first healthcare innovation to UCLA at the age of 14. But even he is surprised by the speed of the digital shift in healthcare. He said COVID-19 accelerated the transition to virtual care by about a decade.
"I think the next couple of years will be, I'll call it, 'multimodal data integration' — it's who you are, it's where you live, it's the experiences that you have — all integrated through algorithms to keep you healthy," he said. "And novel data sources — including the devices you wear, devices in your home, devices you carry — will become an increasingly important part of us keeping you healthy."