Penn researchers foster dialogue about personalized medicine in mental health

In an ideal world, physicians consider factors such as a patient’s cognitive and psychological ability, brain activity, and even environmental stressors to decide the best treatment option. Some cancer care successfully employs this personalized medicine approach, and Penn psychologists Robert DeRubeis and Zachary Cohen are currently working with colleagues around the world to integrate it into mental health.

DeRubeis
Robert DeRubeis, the Samuel H. Preston Term Professor in the Social Sciences in the School of Arts & Sciences.

DeRubeis, the Samuel H. Preston Term Professor in the Social Sciences, specializes in depression research, so that’s where the pair began. They’ve since added other focus areas like PTSD and obsessive compulsive disorder.

In 2014, they published a paper in PLOS One on their Personalized Advantage Index (PAI), a model that provides expectations about a patient’s success with two care plans and the difference between predictions. Though the PAI is still being tested, the scientists showed that when comparing antidepressants to cognitive behavioral therapy for depression patients, the index offered a small but significant decision-making boost. 

Recently, Cohen, a doctoral candidate in the Department of Psychology, partnered with DeRubeis to organize two days of discussion about the field’s latest methods, findings, and challenges. Mental health and cancer researchers, statisticians, geneticists, and policymakers from eight countries participated.

“A lot of people around the world are excited about this,” DeRubeis says, “figuring out how we can move [away] from a one-size-fits-all treatment approach.”

The Penn researchers concentrate on a clinician’s initial recommendation, and consider questions including: Does the practitioner begin with the more or less intensive option? What about when two valid choices arise or when a previously prescribed treatment doesn’t work? The challenge becomes determining which factors—genes or brain imaging or demographic information, for example—should enter into the decision-tool equation and how best to combine that information.

“There is this sense that we can do better [determining factors and making treatment decisions],” DeRubeis says. “That’s not in contrast to but separate from the idea that we need to find new treatments. People are still trying to do that, but many of us think we shouldn’t be counting on that as our savior. Instead, we should be doing the best with the treatments available.”

Getting there will likely mean collaborating, which the Penn researchers already do (and will do more of), thinking creatively about new datasets like those from electronic health records, and conducting careful testing before rolling out these models.

The field is still answering some big questions: How confident in these systems must researchers be before practitioners can use them? How much of an advantage must they confer?

“These systems would only need to provide a 5 or 10 percent benefit,” DeRubeis says. “Given how little they would cost, that would be tremendous.”

Despite some looming concerns, personalized medicine could penetrate mental health within the next five years, according to the researchers. Cohen describes a large clinic in Germany already using two decision-support tools.

“People are talking about big data,” DeRubeis says. “It’s exploiting information that seems invisible, and pulling out [and translating] the signal that can allow for a more potent, rational, efficient delivery of health care.” 

Originally published on .