In 2019, wearable technology is ubiquitous.

Whether used for tracking sleep patterns, monitoring heart rate or logging exercise, nearly 57 million U.S. adults — roughly a quarter of the adult population — will use a wearable device such as a fitness tracker or smartwatch at least once a month this year, according to an eMarketer study. Moreover, wearables are becoming popular among Americans 55 and older. Their use is projected to increase more than 15 percent since 2018.

Noting the upswing in wearable technology use and Americans’ declining activity levels as they age, Ekaterina Smirnova, Ph.D, assistant professor in the Department of Biostatistics in the VCU School of Medicine, and co-leaders of Johns Hopkins University’s Wearable and Implantable Technology group — Ciprian Crainiceanu, Ph.D., and Vadim Zipunnikov, Ph.D. — studied how physical activity levels can predict death.

VCU News sat down with Smirnova to discuss their work. 

Why and how did you conduct this study?

Right now there is general interest in technology and human health, specifically, wearable technology.

For this study, I worked with a group at Johns Hopkins University that included multiple academic and industry researchers in biostatistics, engineering, epidemiology and other fields. Our research was supported by the National Heart, Lung, and Blood Institute; the National Institute of Neurological Disorders and Stroke; and the National Institute on Aging. We are a diverse and collaborative group, and we set out to study mortality levels based on activity in large population studies. In particular, we were looking at five-year mortality in older adults, or, how likely they were to be alive in five years.

Death — and the factors that contribute to it — have been studied before. It’s an important subject. Previous studies looked at association, not prediction. For example, we know that age and activities are associated with death. But in this study, we are actually putting a number on people’s risk, given their activity levels, to predict death.

Overall, we combined the assessment of wearable devices, identified death prediction factors and used quantification to rank how each factor is predicting death. We studied the data of participants ages 50 to 85 over the course of two years, which was available in the Centers for Disease Control and Prevention’s National Health and Nutrition Examination Survey, or NHANES. We also conducted an observational study where we monitored a large number of people — 3,000 participants — ages 50 and older, who represent the diverse U.S. population as much as possible.

What did you find?

When you think of what factors into someone’s death, the first thing you think about is age. There are of course other risk criteria such as the presence of certain diseases, obesity and diabetes that we also consider risk factors for morality. But in our study we found that the best predictor of someone’s five-year mortality was actually their physical activity levels.

Your physical activity level is in some ways a snapshot of your general health. When you go to the doctor, the typical things that are measured are blood pressure, height, weight, cholesterol level and history of chronic diseases. As it turns out, a person’s activity level contains measurements of all these factors together. If a person is sick or has a certain disease, they move slower and take longer breaks because they get tired more frequently. Healthy people and younger people tend to move more. The group previously developed several types of activity measures that take into account all of the aforementioned factors, and collectively, they overwhelm traditional measures in assessing someone’s health.

Can someone predict when they will die based on data from their own wearables?

In principal, yes. It’s a little more complex than that, but we did essentially create a risk calculator to quantify an older adult’s risk of death. It would require entering information like age, weight and other measurements into an algorithm and it would calculate mortality.

Our study is replicable and our analysis is transparent, which is extremely important for a scientific paper. Anybody who is interested, and has general knowledge of our programming language, can follow all of the steps we have in our paper from beginning to end using an analysis vignette that I put together with Ph.D. candidates Quy Cao of the University of Montana and Andrew Leroux of Johns Hopkins University. We include everything from links for downloading data to figures and helpful tables. It’s publicly available online as a part of software designed by Leroux and Jacek Urbanek, Ph.D., also of Johns Hopkins, who made it possible to transition from “available data online” to  “easily accessible and usable data.”

What is the benefit of using data from wearables in health research?

A number of studies assess activity-use questionnaires. The problem with this is that it is subject to recall bias and it’s extremely subjective from one person to another. For example, if a person is asked how much they move, they may think they move a lot but that may differ from their actual activity level.

This particular study activity was measured using accelerometry, which comes from wearable devices similar to Apple Watch or Fitbit. These devices can be worn on people’s wrists, on the hip, sometimes on the ankle — different places on the body. It essentially helps researchers measure the activity levels more accurately.

As we note in our paper, wearable technology is important in providing reproducible, unbiased and prognostic biomarkers of health. If you’re efficiently organized and use consistent data, the results are more reliable.

How do you hope your findings — and using data to measure health — will help people?

Age is something one cannot change. However, activity is a modifiable factor and something you can really change. It would be nice to have activity assessment be part of the routine practice when people go to the doctor.

When you go to the general practitioner, they ask you how much you exercise or move. But that goes back to the same subjective questionnaire I mentioned, and we know there is greater benefit in having objective measurements. They’ll tell you to move more, but it’s difficult to know exactly how much more or at what level you even started. We hope that results of our research will translate to medical practice. Crainiceanu, the founder of the wearables group and senior author on this paper, envisions that doctors will routinely use this technology and take the idea that we can quantify your activity — or put information into a calculator — so they can truly understand your health. Then they can see how much you improve, say if you move 10 percent more over last year.

This is being done to a certain extent. You’ll see people walking around with Fitbits. But what they don’t do is really compare their progress over a long period of time, nor do they consider major life-changing events that affect activity levels — things like graduating from school, getting a new job that’s sedentary, starting a family, and more.

Being able to have this as a part of clinical practice would help people become more self-aware. We’re all very busy and, nowadays, lead fairly sedentary lives. We take ourselves to work in cars and don’t necessarily have time for the gym. We know physical activity is important, but it's very hard to build it into our lives.