Validity of Optical Heart Rate Measurement in Commercially Available Wearable Fitness Tracking Devices
BackgroundWearable fitness tracking devices have risen in popularity for athletes and the general population and are increasingly integrated into smartwatch technology. Many devices incorporate optical heart rate (HR) measurement by photoplethysmography which provides data used to monitor and track exercise training intensities, progress, and other health and fitness related parameters. ObjectiveTo determine the validity of optical heart rate measurement in three fitness tracking devices while resting, walking, and running. MethodsTwenty subjects (10 male, 10 female) completed the research study based on the ANSI/CTI standards for physical activity monitoring of heart rate under 4 different conditions: sedentary (SED), treadmill walking (WLK), running (RUN), and dynamic running/walking (DYN). Subjects wore 3 optical heart rate devices: Polar OH1 (OH1) on the right forearm, Apple Watch 4 (AW4) on the right wrist and Garmin Forerunner 945 (FR945) on the left wrist. A Polar H10 (H10), a chest strap device, was the criterion HR measurement device. SED, WLK, and RUN were all 7-minute protocols with 1 minute of standing, 5 minutes of prescribed activity, and 1 final minute of standing. The DYN protocol was a 12-minute protocol with 1 minute of standing, 10 minutes of variable intensity walking and running, and 1 minute of standing. Raw HR data was extracted from each device and temporally aligned with the criterion H10 HR data for analysis. ResultsThe mean absolute deviation (MAD, measured in beats per minute) for the three experimental devices (OH1, AW4, FR945, respectively) for SED was 1.31, 1.33, and 2.03; for WLK was 2.79, 2.58, and 5.19; for RUN were 4.00, 4.29, and 6.51; and for DYN was 2.60, 2.44, and 2.44. The mean absolute percent error (MAPE) for the three experimental devices (OH1, AW4, FR945, respectively) for SED was 1.78%, 1.89%, and 2.81%; for WLK was 3.15%, 3.18%, and 5.93%; for RUN was 3.43%, 3.51%, and 5.25%; and for DYN was 2.05%, 1.95%, and 5.47%. The intraclass correlation for each device across all conditions was .991 (OH1), .984 (AW4), and .697 (FR945). ConclusionsAt rest, and during both steady-state and variable-speed treadmill walking and running, the Polar OH1, Garmin Forerunner 945, and Apple Watch 4 optical HR monitors demonstrated a level of accuracy well within that required by the ANSI/CTA Standard (2018) for physical activity monitoring devices for heart rate measurement (i.e., <10% Mean Absolute Percent Error). Therefore, consumers can have confidence that these devices provide HR data with accuracy that conforms to the performance criteria recommended for consumer electronics.