Focus on new faculty: Yonatan Mintz makes machine learning work for personalized healthcare

// Industrial & Systems Engineering

Photo of Yonatan Mintz

Yonatan Mintz

Leave it to an optimization researcher who specializes in healthcare to use county-level COVID-19 data to meticulously plot out a move from Atlanta to Madison, Wisconsin.

That’s precisely what Yonatan Mintz did ahead of his drive north to join the Department of Industrial and Systems Engineering (ISyE) at the University of Wisconsin-Madison in late summer 2020.

Mintz, who joins ISyE as an assistant professor after two years as a postdoctoral fellow at Georgia Tech, applies optimization and machine learning methods to tailor healthcare interventions to individuals. He says ISyE’s strengths in optimization and health systems engineering, paired with its long track record of collaboration with medical researchers across the UW-Madison campus, makes it an ideal home for his work.

Mintz’s research portfolio includes leveraging patient data to hone personalized health and wellness solutions through wearable technology, to refine drug dosing plans in intensive care units, and to better model the different variations of Parkinson’s disease.

“The thing that I find interesting is how do machine learning and optimization impact people? How can we use them to impact people for good? How do we understand the negative effects they have?” he says. “I feel like this is where I can make the most impact: these problems of human-sensitive machine learning and optimization and making sure that these algorithms work for us instead of the other way around.”

Born in Israel to parents who ran a startup biotechnology company, Mintz discovered industrial engineering through an engineering competition as a high school student after the family moved to New Jersey. Upon graduating with his bachelor’s degree from Georgia Tech, he joined Caterpillar as a supply chain engineer at the company’s South Milwaukee facility, figuring he would carve out a career in the business world.

While he says he quickly discovered his future wouldn’t include an MBA, he plans to draw on his industry experience when teaching a supply chain management course for undergraduate students in future semesters.

Mintz also worked at Google in data science before pursuing his PhD at the University of California, Berkeley, giving him a glimpse into the potential of machine learning and a thirst to explore how those techniques could benefit society. At Berkeley, he developed the algorithm behind a fitness app that learns users’ exercise and food preferences and then creates goals suited to them.

Mintz and his collaborators partnered with the University of California, San Francisco, School of Nursing to conduct a randomized controlled trial and found that their app outperformed Fitbit in spurring users to exercise more.

“The right motivation, the right goals to get people to lose weight, exercise or decrease their risk of diabetes really depends on that individual person,” he says. “It doesn’t make sense to do a one-size-fits-all approach.”

At UW-Madison, Mintz plans to continue creating models to promote health and wellness via mobile apps and wearable technology. He’s also developing an algorithm to inform individualized dosing strategies for the blood thinner heparin while accounting for differences in how patients metabolize the drug—a particularly important consideration in intensive care units, where patients are confined to their beds and blood clots are serious threats. His work on Parkinson’s disease involves both a model to predict different manifestations of the disease and methods to more clearly explain the factors behind a diagnosis.

Beyond specific applications, Mintz is also interested in finding technical solutions to combat bias—in the data itself or in those working with it—in automated decision-making. It’s the kind of consequential, multilayered issue that inspires him.

“I like doing math, I like helping people and I feel like being a part of UW and ISyE is one of the best positions I can be in to do both of those things,” he says.

Author: Tom Ziemer