JMIR Publications

ABSTRACT

Background:

Fitness trackers and smart watches are frequently used to collect data in longitudinal medical studies. They allow continuous recording in real life settings, potentially revealing previously uncaptured variabilities of biophysiological parameters and diseases. An adequate device accuracy is a prerequisite for meaningful research.

Objective:

This study aimed to assess the heart rate recording accuracy in two previously unvalidated devices: The Fitbit Charge 4 and Samsung Galaxy Watch Active2.

Methods:

20 healthy adults (24.2 ± 4.6 years) performed a study protocol consisting of 5 sedentary, 2 low-intensity and 3 high-intensity exercise phases, in total lasting on average 19 minutes. Participants wore two wearables simultaneously during all activities: The Fitbit Charge 4 and Samsung Galaxy Watch Active2. Reference heart rate data was recorded using a medically certified Holter ECG. Data of reference and evaluated devices were synchronized and compared at one-second intervals. Mean, Mean Average Error (MAE), Mean Average Percentage Error (MAPE), Lin’s Concordance Correlation Coefficient (CCC), Pearson Correlation Coefficient (PCC) and Bland Altman plots were analyzed.

Results:

Overall and across all activities, the Fitbit Charge 4 slightly underestimated heart rate, while the Samsung Galaxy Watch Active2 overestimated it (-1.66 bpm/3.84 bpm). The Fitbit Charge 4 achieved a lower mean average error during sedentary activities (Seated Rest: 7.8 vs. 9.4, Typing: 8.1 vs. 11.6, Laying down (left): 7.2 vs. 9.4, Laying down (back): 6.0 vs. 8.6, Walking slow: 6.8 vs. 7.7bpm), while the Samsung Galaxy Watch Active2 performed better during and after low- and high intensity activities (Standing up: 12.3 vs. 9.0, Walking fast: 6.1 vs. 5.8, Stairs: 8.8 vs. 6.9, Squats: 15.7 vs. 6.1, 9.6 vs. 5.6 bpm). For one participant, the Samsung Galaxy Watch Active2 reported incorrectly high heart rate readings (160-200bpm for approximately 10 minutes), negatively impacting the metrics and leading to higher Bland Altman Limits of Agreement (LoA).

Conclusions:

Device accuracy varies by activity. Overall, both devices achieve a mean average percentage error just below 10%. They are thus considered to produce valid results based on limits established by previous work in the field. Neither device reaches sufficient accuracy during seated rest and keyboard typing.