Harnessing Behavioral Science to Bridge Gaps in Clinical Research

Despite decades of progress in developing new drugs, diagnostics, and technologies that have the power to improve and even save patients’ lives, one persistent challenge remains: getting people to use them. Whether it’s taking a prescribed medication, showing up for a screening, or following through on a lifestyle change, the success of any medical intervention ultimately depends on human behavior.

This is where behavioral science comes in to help us understand and influence that step, according to Niteesh Choudhry, MD, PhD, professor of medicine at Harvard Medical School, professor of health policy and management at Harvard T.H. Chan School of Public Health, and executive director of the Center for Healthcare Delivery Sciences at Brigham and Women’s Hospital. He also serves on the faculty of Harvard Medical School’s Global Clinical Scholars Research Training certificate program. 

Choudhry points out that by applying insights from psychology, economics, and decision science, researchers and clinicians can better understand what drives behavior and can use these insights to design low-cost, scalable interventions optimized for real-life conditions. This can result in stronger adherence, more effective care delivery, and better outcomes across populations.

Small Gains in Medication Adherence Save Lives

For the past two decades, Choudhry has worked with large health systems and insurers to design and evaluate novel strategies to enhance medication adherence and improve the quality of prescribing for common health conditions, such as heart disease and diabetes. Over the course of his research, he has learned that improving health behavior is both essential and challenging, and one approach doesn’t fit all situations.

“If half of people don’t take their medications as prescribed, that’s not abnormal. It’s just life getting in the way. Most people would take them under the right circumstances,” he explains.

For instance, in one large trial published in the New England Journal of Medicine, Choudhry and his team partnered with a health insurer to offer free post-heart attack medications for 6,000 patients, and the results were striking. 

“We improved adherence by five percentage points. That may sound small, but it was enough to reduce heart attacks, strokes, and deaths,” he stresses. “Little wins help. If I can move you even a little, that can have a big impact.” He adds that while the research was conducted more than a decade ago, the results remain relevant.

Moving Toward Precision Behavioral Science

Choudhry believes that the future lies in highly personalized, scalable interventions. “There may not be one magic bullet,” he says, “but there may be five, and each of us may need a different combination.”
In a 2024 study involving patients with diabetes that was published in npj Digital Medicine, Choudhry and his colleagues tested how different types of text messages influenced behavior. “Some people responded to positive messages such as ‘Great job!’ Others preferred the drill sergeant approach with messages like ‘Don’t forget your dose!’ Some needed reminders tied to family, while others responded to statistics,” he says.

To adapt messages to individual preferences, the team used reinforcement learning, a machine learning technique. “We changed the message based on what each person was predicted to respond to, and we saw significant improvements in adherence,” notes Choudhry.

This approach really gets to the heart of what makes each person tick. “It’s about how people think, what motivates them, and how their environment affects their choices,” he says.

The Promise of AI

With the growing popularity of artificial intelligence, Choudhry sees even greater potential. One promising direction involves large language models, a type of language model trained on large amounts of texts so that it “learns” how to generate natural-sounding language.

In the diabetes study, his team created a static bank of messages. While there are limitations in these pre-set messages, he highlights the opportunities to tailor them in new ways.

“Imagine using a large language model to generate context-specific messages,” he says. “If it’s raining, the message might say, ‘It’s a gloomy day. Don’t forget your medication before you settle in with a book.’ That kind of contextualization could make interventions feel more human and more relevant.”

Of course, integrating AI into care raises technical, ethical, and regulatory questions. But Choudhry believes that the potential is enormous. “What would take a human coach hours, AI could help us do efficiently, and at scale.”

Behavioral Science as Discovery Science

For clinical researchers, behavioral science offers a new lens on discovery. “We often think of discovery science as happening in a lab. But understanding human behavior is equally scientific,” Choudhry says.

He calls this emerging field “precision population health,” describing it as the study of how to target interventions to the right individuals and communities using rigorous, multidisciplinary methods. 

“The techniques come from psychology, economics, machine learning, and clinical epidemiology,” he says. “Bringing them together is what makes behavioral research so powerful.”

The key is to remember that health behaviors don’t happen in isolation. “Medication taking, food choices, and exercise all happen in context,” Choudhry says. “And that context includes social factors and life’s chaos.”
Understanding these dynamics is essential. “It’s not about blaming patients; it’s about recognizing barriers. If we can ethically capture contextual data and use it to inform care, we can design smarter, more compassionate solutions,” he says.

Advice for Emerging Clinical Researchers

For clinicians and scientists interested in integrating behavioral science into their work, Choudhry offers two key pieces of advice.

First, recognize that it’s an actual scientific discipline. “Behavioral science and implementation research have established methodologies, frameworks, and a rich literature,” he says. “If this resonates with you, learn more because there’s a whole discipline dedicated to turning evidence into action.”

Second, start with a problem that matters to you. “Each health behavior is different. Exercise isn’t the same as smoking or getting vaccinated,” he says. “Find a challenge that compels you, then borrow the tools you need from across disciplines.”

Choudhry describes himself as “a bit of a clinical trialist, a bit of an epidemiologist, a bit of an economist, a bit of a statistician, a bit of a machine learner . . . and certainly an internist.” He encourages others to take the same multidisciplinary approach. “You don’t need to be a pure behavioral scientist. Just take what you need to solve the problem.”

Turning Scientific Discovery into Real-World Impact

By combining clinical insight with data-driven behavioral strategies, researchers can design smarter trials, build stronger evidence, and deliver care that aligns with how people actually live.

“The more we understand what drives behavior, the more effectively we can close the gap between what we know and what we do. And that’s the key to turning scientific discovery into real-world impact,” he stresses.