Statistician’s journey to ‘genius’ took a lot of hard work

Alumna Susan Murphy is developing smart phone apps to help people who are battling addiction or mental illness. (photo courtesy of MacArthur Foundation)

Alumna Susan Murphy is developing smart phone apps to help people who are battling addiction or mental illness. (photo courtesy of MacArthur Foundation)

If Susan Murphy’s story tells us anything, it’s that “genius” takes a lot of hard work.

Murphy (Ph.D. ’89) is a 2013 recipient of a MacArthur Foundation “genius grant.” The grants provide unrestricted funding to some of the best and brightest in a wide range of fields, from science to the arts, culture and public policy.

The fellowship, accompanied by a $625,000 prize, was awarded to Murphy for her work in using advanced statistical techniques to help find better treatments for mental illness and other chronic conditions.

As the H.E. Robbins Professor of Statistics and a professor of psychiatry at the University of Michigan’s Institute for Social Research, she’s working with psychiatrists, computer scientists and other experts to develop smart phone apps that would help people who are battling addiction or mental illness stay sober and recover.

But it began with a lot of hard work in Chapel Hill.

She came to Carolina in the 1980s as a graduate student in statistics. In her first year she found herself in an academic boot camp. Though she had always been good at math, graduate statistics was exponentially more demanding. “It was extremely intense,” she says. “I just lived, breathed statistics.”

“I would go to class and I would understand the first 10 minutes or so,” she says. “The rest of the class I would just struggle to take accurate notes.”

Murphy discovered, though, that she could go home, work through the notes, and eventually understand everything that had been covered in class. Later, that same approach allowed her to delve into other academic specialties, such as psychology, and work with researchers to apply advanced statistics to their research.

Murphy’s sister, Nancy Allbritton, the Debreczeny Distinguished Professor of Chemistry at UNC and chair of the UNC/NCSU joint department of biomedical engineering, says the two of them are “a bit stubborn.”

“Once we decide to do something, we just do it,” says Albritton. The sisters — two of five girls, all now high achievers — grew up in rural Louisiana. Their parents and grandparents, though, simply expected they would do well in school.

At Murphy’s first stop after grad school, as a statistics professor at Penn State, she reached out to help a colleague in psychology. The psychologist was studying the impact of various therapies on children with behavior problems and was struggling to understand why it appeared that tutoring might be having a negative impact on students’ behavior. It took Murphy three months to figure out what she needed to know — and what she didn’t need to know — before she could help the psychologist with her data.

“Some of what she wasn’t telling me was really important to know,” Murphy says. “Neither one of us knew what I needed to know and what I didn’t need to know.”

Starting with that work, Murphy has delved deeply into how researchers and clinicians implement and understand clinical trials for chronic conditions, which often involve changes in therapy over time.

In the case of traditional clinical trials for acute conditions — think hypertension, for instance — doctors might compare the effectiveness of one drug vs. another. It’s treatment A vs. treatment B.

But in chronic conditions, such as depression or substance abuse, doctors and therapists might implement a series of treatments: first treatment A, then treatment B, then treatment C and so forth.

Clinicians need to assess the effectiveness of each treatment and when it might be time to make a change. Murphy’s work helps researchers run clinical trials that produce those kinds of answers.

Now she’s working with computer scientists and clinicians to develop interventions that might change over time, based on feedback from patients and their needs at a particular stage of recovery. Those smart, adaptive interventions could even be programmed into a smartphone application that could help people better manage their own condition.

The work has many applications beyond chronic mental illness. Potentially, it could be applied to everything from weight loss to making sure people stay on maintenance medications necessary for their health.

“This is to me the big open area,” Murphy said. “How can we use data to help us figure out the best way for patients to either manage themselves better, or help clinicians figure out how to manage them better?”

[ By Mark Tosczak ]