11 minute read

WEARABLE TECHNOLOGY

On the threshold of clinical care

By Craig Collins

n IT’S A STRANGE-SOUNDING NAME for a medical outfit, but the Penn Medicine Nudge Unit is just that: a unit that nudges. Director Mitesh Patel, M.D., MBA, M.S., and colleagues at the University of Pennsylvania’s Perelman School of Medicine design interventions that align patient behaviors and decision-making with long-term health goals: making the right choices obvious to patients, and making those choices easier to select.

As a primary care physician at the Corporal Michael J. Crescenz Veterans Affairs Medical Center in Philadelphia, Pennsylvania, Patel realizes that nudging patients – many of whom he sees only once or twice a year – toward healthy behavior is no simple task. “One of the challenges we often have,” he said, “is that we’re talking to patients about how they should be more physically active or lose weight, and about changing their behavior. But then they leave our office, and we don’t see them for six months or a year – or some longer period. And we don’t have any way to interact with them outside of the visit.”

Patel’s recent research has focused on monitoring patient activity and weight – often with the use of consumer fitness wearables from manufacturers such as Apple®, Fitbit, or Samsung – combined with incentives for behavior change. “In most cases,” he said, “just giving someone a device and a smartphone is not effective in changing their behavior. But if you combine it with the right behavior change strategy, it can be really effective.”

In recent studies Patel led for the University of Pennsylvania, his team issued activity monitors to recently discharged heart patients and monitored their progress toward a target goal of activity. His studies have used both financial incentives – setting up a modest account and subtracting money when goals aren’t met – and “gamification” incentives, aimed at families who earn or lose points based on whether family members meet their goals.

His studies have shown that these structured behavior modifications, followed closely with wearable activity monitors, can sustain behavior changes for up to six months, a vast improvement over patients who are simply given the monitors. “There’s a lot of good evidence to show that half of the people who get a wearable device stop using it within a couple of weeks to a month,” Patel said. He’s in the process of launching his first study to compare the effectiveness of these incentives among veterans who are overweight or obese, with a body mass index (BMI) of 25 or greater. His team will track step counts and minutes of moderate to vigorous physical activity among groups of veterans working with each incentive scheme, and compare outcomes.

Step counts and rates are fairly simple data for measuring physical activity, but Patel said that among consumer wearables, those are the only two measures he trusts for his purposes. “We’ve found they’re fairly good at tracking step counts,” he said. “But other studies have found that their monitoring of other things, like sleep and heart rate and calories, is actually not that great. The technology isn’t there yet. You have to really trust the data in order to make use of it.”

It’s one of the biggest issues facing wearable technology today: Stories abound of wearable technologies on the threshold of revolutionizing treatments for certain diseases or disorders, but we’re not there yet, particularly in the area of medical-grade wearables designed to inform clinical decision-making. There are a few at work today: The Zio® wireless patch, a small adhesive patch that can be worn on the chest to monitor heart arrhythmias for up to two weeks, was cleared in 2011; four continuous glucose monitors (CGMs) have been cleared by the federal Food and Drug Administration (FDA) for use in monitoring diabetes. Last year KardiaBand, manufactured by AliveCor, became the first medical device accessory approved for the Apple Watch by the FDA, and several other devices – including a blood pressure monitoring smartwatch, a sleep apnea-monitoring Fitbit, and a wristband that will monitor a wider range of biometric information such as blood oxygen saturation and respiratory rate – are on the near horizon.

An Apple Watch®. In 2017, the KardiaBand became the first medical device accessory approved for the Apple Watch by the Food and Drug Administration. The device is a clinical-grade wearable EKG band that replaces the Apple Watch band, providing access to an EKG anytime, anywhere.

An Apple Watch®. In 2017, the KardiaBand became the first medical device accessory approved for the Apple Watch by the Food and Drug Administration. The device is a clinical-grade wearable EKG band that replaces the Apple Watch band, providing access to an EKG anytime, anywhere.

FANCYCRAVE1 VIA WIKIMEDIA COMMONS

In the meantime, Department of Veterans Affairs (VA) researchers continue to explore how wearable devices might be used to improve patient care. At the VA’s Advanced Platform Technology (APT) Center in Cleveland, Ohio, clinicians and IT engineers work together to study how remote sensing can be used in rehabilitative medicine. Applications currently being developed by teams at the APT Center include:

• A wearable sensor, part of a smoking cessation intervention, that captures hand-to-mouth smoking motions and is sophisticated enough to differentiate them from other movements (i.e., answering a phone call or drinking). The sensor cues the sending of tailored intervention video content to the relapsed smoker.

• A customizable cloth-like pressure sensor that can be inserted between the end of a residual amputated limb and a prosthetic. The sensor maps the pressure distribution, sends that data to a visualization system, and allows for prostheticians to redistribute forces that might lead to skin ulceration.

• A “smart” wheelchair cushion that senses how hard and how long pressure is distributed over itself. With the use of algorithms that analyze pressure data, the cushion will alter its shape and stiffness in response.

• A wearable gait laboratory, an insole array of sensors that monitors slips and trip and falls in everyday environments through analysis of gait parameters, balance controls, and body postures.

APT investigators include Case Western Reserve University faculty members Rahila Ansari, M.D., M.S., an assistant professor of neurology, and Ming-Chun Huang, Ph.D., assistant professor of electrical engineering and computer science. Ansari is also a practicing neurologist at the Louis Stokes Cleveland VA Medical Center. According to Huang, a technology such as the wearable gait laboratory may be a useful tool for clinical evaluation, whether the data is gathered in the hospital or at home. “With the research and the data, we’re gathering,” he said, “we’re trying to accumulate knowledge to see how wearables might trigger effective interventions after an injury.”

A Zio XT monitor in place on a woman’s chest and the Zio XT monitor. The wireless patch can be worn to monitor heart arrhythmias for up to two weeks.

A Zio XT monitor in place on a woman’s chest and the Zio XT monitor. The wireless patch can be worn to monitor heart arrhythmias for up to two weeks.

COURTESY IRHYTHM TECHNOLOGIES, INC.

An important distinction between many APT projects and most of the wearables available today, said Ansari, is that APT investigators hope to design “closed-loop” systems that take measurements, transmit that data into a control system that analyzes it using algorithms, and activate a mechanism that can take the necessary action. “For example,” she said, “if we’re dealing with a prosthetic limb or a wheelchair, then in order to prevent a breakdown or ulceration, we measure what the forces are, and then we’ll continuously adjust for all those things in real time, so we know we’re preventing problems.”

CLOSING THE LOOP: ALGORITHMS AND ANALYTICS

It’s these next-generation wearable sensing technologies – not devices but systems, of which devices are a component – that signify a transformation in the way doctors and patients interact. Their artificial intelligence will indicate not only what the data say, but what should be done in response.

Josef Stehlik, M.D., MPH, a professor at the University of Utah School of Medicine and cardiologist at the VA Salt Lake City Health Care System, specializes in patients with heart failure, a condition that involves high readmission rates among patients discharged from hospitals. In spring 2018, Stehlik’s research team reported the results from their study of a wearable monitoring system: “a Band-Aid-like patch,” he said, “containing several sensors that detect patients’ physiological parameters.” One hundred veteran patients from four different VA medical facilities wore the monitors, which transmitted data using Bluetooth technology to their smartphones and tablets – which, in turn, uploaded the data to a secure VA server in Sacramento, California.

From there, data from Stehlik’s 100 veteran heart patients, including parameters such as heart rate, respiratory rate, posture, and activity, were fed into an algorithm that compared them to a previously established baseline for each patient. “This predictive algorithm was very accurate in identifying which patients were likely to get in trouble with heart failure exacerbation,” said Stehlik. For patients whose data were heading into dangerous territory, the analytics triggered an alarm that notified the research team. “We’ve also shown that this alarm would come approximately seven to 12 days before the readmission would happen,” Stehlik said. “So presumably, there would be sufficient time to do an intervention: contact the patient, change medications to treat the patient before the exacerbation progresses, on time to prevent a readmission.”

Now that his team has established the ability of the analytics to predict exacerbation for heart patients, Stehlik hopes to show how the data can be integrated usefully into cardiologists’ clinical workflow. “As you can imagine, a lot of clinicians have been bombarded by lots of different data. It’s not just important to make data available – it needs to be processed into an output so that clinicians can respond to it and provide the benefit of that information to the patient. That’s where I think a lot of research is necessary.” He hopes to conduct a trial among VA patients in which he can measure the clinical efficacy of the wearable monitoring system: whether the device paired with a predictive algorithm can reduce readmissions or shorten hospital stays for heart patients.

A graphic depicting an interpretation of the “Internet of Wearable Things.”

A graphic depicting an interpretation of the “Internet of Wearable Things.”

IMAGE COURTESY OF DR. MING-CHUN HUANG, EECS CWRU

At the Daroff-Dell’Osso Ocular Motility Laboratory at the Cleveland VA Medical Center, neuroscientist Aasef Shaikh, M.D., Ph.D., a professor at Case Western Reserve University, is examining how sensors can be used to aid in diagnosing and differentiating among different tremor disorders that often present similarly. In the laboratory, Shaikh and his colleagues fit patients with sensors – wearable magnetometers with gyroscopes capable of taking fine 3-D positional readings. Just as with Stehlik’s heart patients, this data has been fed into specifically formulated algorithms to distinguish one type of tremor – for example, cervical dystonia, essential tremor, or Parkinsonian tremor – from another.

These assessments currently happen in Shaikh’s lab, but he and his team are developing mobile versions of the sensor that will be able to transmit data wirelessly. “The key thing here,” he said, “is that now we can use wearable technology and machine learning, artificial intelligence, to analyze the output of those wearables to provide better diagnostics – and better care – for the patient.”

Shaikh’s ambition is to develop a closed-loop system that may become directly involved in providing care for tremors. Deep brain stimulation (DBS), a method of dampening muscleactivating signals from the thalamus, is achieved among tremor patients through surgery, in which electrodes are implanted into the thalamus and connected to a generator – Shaikh describes it as a “pacemaker for the brain” – that can block tremor-inducing impulses.

With a wearable sensor that can detect the tremor, and an algorithm that can distinguish the tremor type, the system would be capable of applying DBS to relieve the patient’s symptoms.

THE FUTURE: THE INTERNET OF WEARABLE THINGS, BIG DATA, AND DOCTORS

Closed-loop systems with wearable technology may be the key to offering real-time interventions for patients with burdensome conditions that require constant vigilance. Huang, Ansari, and their colleagues at the APT Center, for example, are looking to close the loop for patients at constant risk of having their skin integrity compromised: patients who wear prosthetics or use wheelchairs. If Shaikh can close the loop for tremor patients, he can offer them a degree of relief they’ve never known.

For nearly two decades, people with insulin-dependent diabetes have been using continuous glucose monitors (CGMs): Wearable sensors that detect glucose levels in the intracellular fluid just under the skin, transmit readings wirelessly to receivers or smartphones, and cue patients to take steps such as adjusting insulin or boosting their blood sugar levels. The CGM loop was closed in the fall of 2016, when the FDA approved the first hybrid closed-loop CGM system, Medtronic’s MiniMed670G, that collects and analyzes glucose data with sophisticated algorithms that determine both the timing and amount of insulin needed by the patient. When the system is placed in auto mode, data is used to trigger a small insulin pump for calculated delivery of insulin.

Brian Layden, M.D., Ph.D., associate professor of medicine at the University of Illinois at Chicago and an endocrinologist with the Jesse Brown Veterans Affairs Medical Center, said the closedloop CGM is useful for people with Type 1 diabetes, the autoimmune form that often requires patients to inject insulin several times daily. “Type 1 diabetes is increasing, but not dramatically. It only represents about 5 percent of our diabetes cases now.” The vast majority of Layden’s patients, particularly older veterans, have the more lifestyle-dependent Type 2 diabetes, which is often treated with oral medications.

This hybrid insulin pump/glucose sensor closed-loop device collects and analyzes glucose data and self-adjusts to keep sugar levels for the wearer in range.

This hybrid insulin pump/glucose sensor closed-loop device collects and analyzes glucose data and self-adjusts to keep sugar levels for the wearer in range.

COURTESY MEDTRONIC VIA WEBSITE

Casual observers may notice an important element missing from this loop: the physician. But when clinicians such as Ansari and Shaikh talk about closing the loop, they’re not talking about making themselves obsolete; they’re acknowledging the impossibility of monitoring and delivering health care around the clock. Layden spends plenty of time with patients – both those with Type 1 and Type 2 diabetes – and is intimately involved in their care. Said Huang: “We don’t want patients to feel as if, once they’re given a technology, they’ll rely on that and never meet with the doctor anymore or do therapy.”

Ansari foresees the next generation of wearable monitors, and even closedloop systems, as complementary to physicians’ care. “My thought as a physician,” she said, “is that it will help improve patients’ relationships with their doctors.” Her migraine patients, for example, often have difficulty documenting symptoms in their prescribed headache diaries. “If there were a way to automatically sense that, or if we were able to find a way to have patients input data with the click of a button, all of the sudden that improves the doctor/ patient relationship and makes it so much easier for us to give the advice they’re going to need to improve their migraines, or whatever else is going on.”

When he looks to the future, Huang, the computer scientist, envisions an “Internet of Wearable Things” that not only assists in the care of individual patients, but also forms pools of data that can be studied and mined by research physicians – essentially combining the data collected by wearables with environmental context to improve clinical care in the same way map applications crunch constant streams of user data to recommend the best routes for other users.

Such an approach, he realizes, raises important questions about the security of patient data, particularly when it’s being processed by an artificial intelligence to perform big-picture analyses. “That makes it more complicated to protect the data itself,” he said. “Of course, some people say we should never share that data, but that contradicts our vision of the Internet of Wearable Things – that we can use that data from sensors to improve other people’s lives.”

When the next generation of wearables begins to affect patient care – and perhaps collect data that can be aggregated for big-picture analysis – this concern will become paramount, Ansari cautions. If doctors and researchers are collecting mountains of data about patients, there will need to be safeguards in place to keep patient-specific data out of the hands of third parties with other motivations, such as employers or insurers. “These are some questions we don’t necessarily have the answers to right now,” she said, “but I think we’re going to have to sort [them] out as time goes on.”

Still, she sees the next generation of wearables as inevitable – and welcome – tools for improving patient care, which is why she’s devoted so much time and effort to investigations at the VA’s Advanced Platform Technology Center. “As these technologies grow and we get even a better understanding of how we’re able to use them,” she said, “I think we’re going to be able to get a lot more holistic and well-rounded care provided to our patients.”