Ruth Carlos, MD, on Bridging Radiology, Population Science, and Real-World Care

Soon after beginning her career as a radiologist, Ruth Carlos, MD, MS, FACR began to focus on a central question: how can care better reflect what matters most to patients while improving outcomes at scale? That question has guided her research across patient-centered outcomes, access to care, and the real-world delivery of clinical advances. 

As co-leader of the Cancer Population Science program at the Herbert Irving Comprehensive Cancer Center (HICCC), Carlos brings this expertise to the program’s efforts to advance cancer research across populations. 

From day-to-day care to population research

Carlos’s path began with a desire to extend her reach as a clinician. As a Robert Wood Johnson Clinical Scholar—one of a small group of radiologists in the program—she became interested in how everyday clinical work could translate into broader impact. 

Carlos initially focused her research on cost-effectiveness and technology assessment but quickly realized that traditional metrics often missed what mattered most to patients. 

“The outcomes that matter to patients aren’t always the ones we predict,” she says. “It might be something as simple as, ‘Will I still be able to pick up my grandchild after treatment?’” 

The unintended consequences of care

Over time, that work led her to a defining focus: the unintended consequences of well-intentioned interventions. 

In one early study, her team developed a tailored intervention to increase HPV vaccination uptake in teenage girls by delivering information to mothers during their own cancer screening visits. The idea made sense in theory, but failed in practice. 

“Women came in focused on getting their screening done,” she says. “They didn’t want to stay longer, even for something important like vaccination for their kids.” 

Other interventions also had unintended consequences. Efforts to address vaccine hesitancy sometimes had the opposite effect, reinforcing resistance rather than reducing it. Policy changes, too, produced mixed results. While they found the Affordable Care Act expanded access to free screening, it did not significantly change adherence rates in most populations. And even small out-of-pocket costs could deter patients from returning for follow-up screenings. 

“These are all examples of things that seem like they should work,” Carlos says. “But if we don’t take the time to understand how care actually happens, we miss what really drives behavior.” 

Connecting biology, behavior, and systems

More recently, Carlos’s work has expanded to explore how social context shapes cancer risk and outcomes—not only through access to care, but through biology itself. 

Her team is investigating how chronic stress and adverse social conditions may activate inflammatory pathways that contribute to diseases like breast cancer. By integrating clinical data, patient-reported outcomes, and biological samples through resources like the Columbia University Biobank, the goal is to better understand how risk develops over time. 

“Not everyone with genetic risk develops cancer,” she says. “So what happens in between? That’s what we’re trying to understand.” 

Bridging imaging, AI, and population science

At Columbia, Carlos also holds leadership roles in the Department of Radiology and the Center for Imaging Biomarkers and Integrated Diagnostics (CIMBID), where she focuses on outcomes and care delivery research, with an eye to integrating artificial intelligence output into care delivery. 

“AI can generate incredibly powerful insights,” she says. “But it only becomes meaningful if we can incorporate it into clinical workflows and translate it into action for patients.” 

That translation gap is a central focus of her work. For example, new imaging models can predict a patient’s future cancer risk, but systems to act on that information, from clinician workflows to patient communication, often lag behind. 

Her research also examines how patients interpret and respond to new forms of data, as well as the ethical and practical implications of integrating AI into care. 

“We’re very good at focusing on the upside of new technologies,” she says. “We’re not as good at anticipating the downside.” 

Building connections across Columbia

Carlos sees Columbia—and the HICCC in particular—as an ideal environment to tackle these challenges, given its depth across clinical care, public health, data science, artificial intelligence, and more. 

“I moved here because there were so many people I wanted to work with at Columbia,” she says. “It's a privilege and a pleasure to now do really interesting, fun work with those individuals.” 

As co-director of the Cancer Population Sciences program, her vision is centered on connection: linking researchers across disciplines, lowering barriers to collaboration, and creating space for new ideas to emerge. 

“There are so many people doing interesting work,” Carlos says. “I want to showcase that in a way that builds connections and decreases the barriers to discovery.”