Enter “smartwatch” into the Amazon search bar and more than 30,000 results appear. Many pair with smartphones to track steps, sleep, and heart rate. Some monitor laps in the pool. Others assess antioxidant levels to help guide healthier eating.

But how accurate is the data these devices generate? And can it truly help people monitor their health?

Lindsey Rosman, a data scientist in the UNC Division of Cardiology, urges users to stay mindful.

“If understanding your health with a wearable device encourages you to get some extra steps, sleep better, and stay healthy — that’s a wonderful thing,” she says. “But if your watch is causing you stress and anxiety, that’s something to pay attention to.”

Rosman has seen that harm firsthand. She and cardiologist Anil Gehi once treated a 70-year-old patient with atrial fibrillation — a chronic irregular heart rhythm — who performed more than 900 electrocardiograms (EKG) on her smartwatch in a single year. Whenever the device returned an inconclusive reading, she’d run test after test, searching for a clearer result.

“Over time, we saw this huge increase in the number of EKGs she was taking on a daily basis,” Rosman shares. “It led to multiple trips to the emergency department, caused conflict with her spouse, and disrupted her quality of life.”

Cases like this are exactly why Rosman’s specific expertise matters. A clinical health psychologist with training in data science, she studies information from heart‑monitoring wearable and implant cardiac devices to determine who is at risk for heart disease, how lifestyle choices shape that risk, and what treatments may help those already living with a heart condition.

“I’m an odd duck,” she says with a laugh. “Maybe a better word is ‘unicorn.’”

From bites to bytes

Rosman’s path to cardiology reflects that blend of disciplines. She began her academic journey at Purdue University, majoring in biomedical engineering. Fascinated by her coursework in biology and neuroscience, she joined a lab studying how amphetamines — stimulants that speed up the body’s functioning — affect the brain.

“I was working with hamsters and one bit me,” she recalls, smirking. “I decided I wanted to stay in this space but have less likelihood of being bit by my subjects. People seemed like a logical next step.”

Upon graduating, Rosman enrolled in the clinical health psychology PhD program at Eastern Carolina University, where she was mentored by a cardiac electrophysiologist specializing in irregular heart rhythms and implanted devices. For her dissertation, she researched heart failure during pregnancy and built a data registry to study this rare patient population.

“That was actually the first registry I built,” Rosman says. “It helped us understand health and quality of life issues for this group — one we previously had little information on.”

After earning her PhD and completing her residency at Brown University, Rosman headed to the Yale School of Medicine for a three-year fellowship in medical informatics and cardiovascular outcomes research. There, she began working with electronic health records and other large data sources, learning how to code and build algorithms to analyze patient data and improve clinical care.

Now at UNC-Chapel Hill, Rosman employs this rare, interdisciplinary toolkit connecting human behavior with complex medical data to tackle some of cardiology’s most pressing questions.

A surge in alerts and anxiety

Atrial fibrillation (AFib) is the most common heart arrythmia, defined by an irregular beat in the heart’s upper chambers. Some people with AFib are symptomless. Others describe the sensation as a sudden flutter in the chest, a skipped beat that jolts them upright, or a heaviness that leaves them lightheaded and struggling to catch their breath.  

For many, that uncertainty is the hardest part. When will the next episode hit? Is it dangerous? Constant vigilance can slip easily into worry — and in the era of smartwatches, worry has a way of multiplying.

Rosman has seen how quickly anxiety can spiral. The patient who ran more than 900 smartwatch EKGs in a single year is a case she still thinks about. Most of those readings were perfectly normal; only 55 even hinted at possible AFib. But the device’s inconclusive alerts fed the patient’s stress levels, pushing her to test again and again. It reached the point where she needed cognitive behavioral therapy just to reclaim her daily life.

As wearable use grows, clinicians feel the pressure too. In a 2024 study, Rosman discovered that one in five AFib patients contacted their doctor after receiving an abnormal rhythm notification. For already overextended clinics, every alert becomes potential triage.

“Wearables are changing our systems of care,” Rosman says. “They lack clinical trial testing, evidence, guidelines, and physician recommendations based on those guidelines.” 

Her work aims to shift that dynamic. Rosman wants patients to understand what their devices can and can’t tell them, and to help clinicians know which technologies are actually reliable. She sees wearable data as a powerful tool, but only when people have the knowledge and support to interpret it wisely.

“Industry professionals, health care providers, researchers, and patients are all part of this ecosystem,” she says. “To make wearables truly helpful, we have to work together.”

But wearables are only part of the story. Implanted cardiac devices like pacemakers and defibrillators offer a different kind of window into the heart — one that’s steady, continuous, and remarkably rich. Some patients have decades of rhythm data stored inside their devices, recording their heartbeats across life events, health changes, and aging.

To Rosman, these vast archives of rhythms and behaviors are an opportunity. Her team is building AI-powered software that can analyze data from both wearables and implants, learning to predict who is at risk for stroke or heart failure and which treatments might help most.

“Then the patient and doctor can sit down together and weigh the risks and benefits of each option,” Rosman says. “Eventually, we’d like to include cost data, because medications can be very expensive.”

Her team is now navigating the patent process for two machine‑learning algorithms — one designed to help prevent stroke, the other to guide heart‑failure care. They’re also leading clinical trials to examine how wearables affect patient psychology and health outcomes. These steps move toward the same goal: a future where data empowers patients rather than overwhelms them.

“We need to do the clinical trials to understand the risks and benefits of wearable devices like we would with any other medical intervention,” Rosman says. “We would never prescribe medication to our patients without knowing who’s going to benefit, who’s at risk, and how to mitigate those risks.”

A record of rhythms

To power this work, Rosman has spent years building the largest cardiac-device registry in the United States — and possibly the world. By combining anonymized patient data from all 11 hospitals in the UNC Health System with information from the U.S. Department of Veterans Affairs, she’s created a research engine capable of revealing patterns no single clinic could ever see.

“And they were built to be parallel registries,” Rosman says. “That means we can develop and test AI and machine-learning tools in one data set, then deploy and evaluate them in another where the patients and health systems are completely different.”

The data Rosman works with is de-identified and locked behind layers of servers and firewalls — a digital vault designed to protect the people behind the numbers. She and her team follow the university’s strict protocols for handling this information, treating each step with care to preserve patient privacy.  

“As investigators, we absolutely have to respect that,” she says.

These registries have opened doors to projects Rosman never imagined at the start of her career. During the 2016 presidential election, for example, she linked the registry with voter records and electronic health data to investigate whether nationwide stress affects heart rhythms.

“We were actually the first group to explore that,” she says. “We wanted to understand the stress of politics and its cardiovascular implications. When we looked across multiple years, we found a 77% increase in arrythmias during the 2016 election compared to other periods.”

The registry became just as valuable during the COVID-19 pandemic. As telehealth replaced in-person visits, Rosman used the data to uncover how many patients were missing critical appointments simply because they didn’t have internet access — a finding that helped health systems rethink their approach to virtual care.

She has even shared the platform with the Environmental Protection Agency, which used it to examine how heat and humidity influence cardiovascular disease. That collaboration has grown into an ongoing study exploring how exposure to forever chemicals like PFAS affects heart health, with Rosman’s registry again providing the backbone of real-world evidence.

Each project points to the same possibility: a future where massive, interconnected data sets help researchers understand not just disease, but the world patients live in — their stress, their environments, their access to care, even their politics. And for Rosman, that future is already taking shape.

“Data science is really starting to become more mainstream in cardiology,” she says. “We’ve been collecting a lot of data for a long time, and it’s just been sitting there. We now have tools to better utilize that data to answer important questions for our patients, clinicians, and health systems.”