Smartwatches have emerged as an increasingly prevalent tool for individuals striving to monitor their health and fitness levels.
Approximately 100 million—nearly four in ten U.S. adults—own a smartwatch, while global market analyses suggest that around 560 million individuals worldwide own these devices.
However, an accumulating body of research indicates that one of the most frequently utilized metrics offered by smartwatches—the number of calories expended—exhibits significant inaccuracies.
As an exercise scientist engaged in the study of physical activity and the intersection of wearable technology, I recently collaborated with colleagues from Florida International University’s Medical Photonics Laboratory to investigate the accuracy of common smartwatches in measuring physiological responses during exercise, and whether variations in accuracy are present among different individuals.
Our findings revealed that these devices can display significant discrepancies in their caloric expenditure measurements, with the error margins amplifying as the wearer’s body fat percentage increases.
Understanding Smartwatch Mechanisms
Most smartwatches incorporate accelerometers to gauge movement, alongside photoplethysmographic sensors to approximate heart rate.
The volume of blood coursing through the wrist alters with each heartbeat, as blood absorbs light; this allows the watch to assess shifts in light reflected back to the sensor. Such patterns enable the device to estimate heart rate.
The smartwatch amalgamates data from these sensors with user-specific information, such as age, gender, height, and weight, subsequently employing an algorithm to estimate energy expended by the individual.
While heart rate and movement are comparatively straightforward signals for a wrist-worn gadget to detect, caloric expenditure entails a more intricate evaluation.
This process necessitates an estimation of something that cannot be directly measured, based on several measurable variables.
Variations in users’ body composition, fitness levels, movement efficiency, and individual physiological responses to exercise can all substantially influence predictions.
Study Methodology
For our investigation, we enlisted 58 Hispanic adults from Miami, where our laboratory is located. The participants displayed a diverse range of body fat percentages and skin tones, enabling us to assess not only the overall accuracy of the smartwatches but also the influences of individual differences on accuracy during testing.
Each participant engaged in a standardized cycling protocol while donning four prevalent commercial devices: the Apple Watch Series 8, Garmin Forerunner 955, Samsung Galaxy Watch5, and Fitbit Sense 2. Participants alternated between moderate and vigorous exercise on a recumbent bicycle for 10 minutes.
During this physical activity, we assessed their energy expenditure with a metabolic analyzer that accurately captures the inhalation and exhalation gases through a mask.
In exercise science, this methodology represents the gold standard for measuring calorie expenditure, providing us with a reference point to compare against the smartwatch estimates.
This experimental design afforded the smartwatches a controlled environment conducive to accurate performance evaluation.
By utilizing recumbent cycling, we effectively managed exercise intensity and minimized wrist movement—an essential factor for maintaining reliable heart rate measurements.
Each participant donned two watches, one on each wrist, with placements randomized to eliminate potential bias. No significant advantage was detected in placement.
Evident Discrepancies
Our analysis revealed pronounced errors across all four smartwatches in gauging caloric expenditure. The median error for these devices ranged from approximately 15% to 25%, indicating that, irrespective of individual variances, the watches frequently provided calorie estimates that were significantly at odds with our laboratory measurements.
Apple, Garmin, and Samsung consistently overestimated caloric expenditure. Apple exhibited the smallest average systematic error, while Garmin and Samsung reflected substantial overestimations.
Conversely, Fitbit presented a distinctive challenge. Despite the absence of consistent positive or negative biases post data-quality exclusions, it yielded several implausible readings—most notably, seven estimates surpassing 450% of the caloric values recorded by the metabolic analyzer.
The general inaccuracy of these devices, while anticipated, has been widely documented in prior studies. What came as a revelation was the pronounced correlation between body fat percentage and measurement errors.
Increased body fat percentages consistently correlated with heightened inaccuracies in calorie estimates across all tested devices, though the extent of this effect varied by brand, with inaccuracies reaching 50% to over 100% in participants displaying elevated body fat levels.
For illustrative purposes, should our esteemed metabolic analyzer identify an energy expenditure of 100 calories during a workout, a 50% overestimation would yield a display of approximately 150 calories on the smartwatch, whereas a 100% overestimate would present as 200 calories.
We determined that skin tone did not significantly impact caloric expenditure errors under the conditions of our study. All four smartwatch brands assessed exhibited considerable errors in caloric measurement.
Implications for Smartwatch Users
This study does not elucidate the reasons behind the substantial impact of body fat percentage on accuracy.
The development and calibration of estimation algorithms likely play a critical role, yet due to the proprietary nature of these algorithms, exact methodologies remain inaccessible.
Individuals with higher body fat percentages are often the very users inclined to leverage wearable devices for weight management goals. Our findings suggest, however, that they may receive the least precise caloric expenditure data.
Errors of 50 or 100 calories during short exercise durations may appear negligible; however, such discrepancies accumulate.
Over an extended period, for individuals with a body fat percentage of 35% or greater exercising four hours weekly, the smartwatch could misrepresent caloric expenditure by an alarming 1,200 to 2,400 calories each week.
This miscalculation could inadvertently lead to consuming a significant excess of calories that one believed were expended through workouts.
Strategies for Consumers
It is important to clarify that the utility of smartwatches is not nullified by these findings.
In a complementary investigation employing the same exercise parameters, our team discovered that these devices exhibit reasonable accuracy in heart rate measurement.
Depending on the activity, they can also provide valuable insights into additional metrics, such as workout duration, distance, and pace.
Based on our research, it is advisable for smartwatch users to regard the caloric estimates as rough approximations rather than precise measurements.
For those aspiring to lose weight, relying on these figures to dictate dietary intake or evaluate workout effectiveness could be ill-fated.
Examining trends over time may prove beneficial, particularly when integrated with heart rate data, exercise intensity, variations in body weight, and personal experience.
Nonetheless, consistency does not guarantee accuracy; a device can consistently present inflated caloric estimates while seemingly maintaining uniformity across multiple workouts.
Advancements in wearable technology are inevitable. Enhanced sensors may contribute to this evolution; however, improving the estimation of energy expenditure across diverse body compositions may hold equal or greater significance.
Within our laboratory, we are expanding our smartwatch experiments to strength training, a domain where the amalgamation of wrist and full-body movements, compounded by varying exercise intensities, may render accurate estimations even more elusive.
Concurrently, research labs worldwide are striving to enhance these technologies and refine their ability to inform health-related decisions more effectively.
Source link: Theconversation.com.
