Author(s): LEE, J.1, PARK, J.1, SON, S.1, LEE, J.Y.1,LEE, H.1, YUN, S.1, JUNG, A.2, KIM, G.2, KIM, H.2, LEE, D.T.1, Institution: KOOKMIN UNIVERSITY, Country: KOREA, SOUTH, Abstract-ID: 2193

INTRODUCTION:
This study was conducted as part of a usability evaluation to support the development and optimization of exercise algorithms in Samsung Health. The purpose of this study was to examine the agreement of heart rate measurements between the Samsung Galaxy Watch and other commercial wearable devices across a range of exercise conditions, using the Galaxy Watch as the reference device, with comparisons performed according to exercise modality.
METHODS:
Five healthy adult participants were recruited for this study. All participants wore a Samsung Galaxy Watch and additionally wore one or two wrist-worn wearable devices selected from six brands (Apple, Coros, Garmin, Suunto, Pixel, and Whoop) on both wrists simultaneously. Participants completed three exercise protocols on separate days, with exercise intensity standardized based on Galaxy Watch measurements: 1) Level walking: walking on flat terrain at an average speed of 5.45 km/h for 1.96 km in 21 min, 2) Resistance exercise: 3 sets of squat, push-up, burpee, and mountain climber with each performed for 1 min per set, 3) Flexibility exercise: 25 mat Pilates and full-body stretching, including elbow plank, roll back, and toe taps for 29 min. All devices were synchronized to start and stop recording simultaneously. After each session, raw heart rate data were extracted and synchronized at 15-second intervals for time-series analysis of real-time fluctuations.
RESULTS:
Analysis of heart rate differences between the Galaxy Watch and other wearable devices revealed exercise-dependent variability. During flexibility exercise, the Galaxy Watch tended to report higher values than other devices, with discrepancies exceeding 20 bpm in some segments. During walking, bias ranged from −15.7 to 14.9 bpm and varied across devices, indicating substantial algorithm-related variability. During resistance exercise, most comparator devices recorded higher heart rates than the Galaxy Watch, reflecting predominant underestimation. Large discrepancies of up to 40 bpm were observed during high-intensity upper-body movements. Measurement stability gradually improved over time during flexibility and walking exercises.
CONCLUSION:
The findings of this study indicate that agreement in wrist-worn heart rate measurements varies substantially according to exercise modality and movement characteristics. Although walking is considered a cyclic activity with relatively low measurement error, considerable inter-device variability was observed, suggesting that differences in proprietary calibration algorithms may serve as a primary determinant of measurement accuracy. Furthermore, during resistance exercises involving substantial upper-body movement, comparator devices tended to overestimate heart rate relative to the Galaxy Watch, highlighting the need for caution when interpreting heart rate data interchangeably across wearable devices.