Author(s): NASSIS, G., MARQUES, E.A., ALQAYDI, A., AHMAD, B., DAAB, W., Institution: UNIVERSITY OF KALBA, Country: UNITED ARAB EMIRATES, Abstract-ID: 1213

INTRODUCTION:
Wearable devices that use photoplethysmography (PPG) to assess heart rate (HR) are widely used but it seems only a few of them have been tested for validity and reliability. Moreover, it is speculated that the ambient light may interfere with the PPG signal, affect its quality and the derived HR values (1). The aim of this study, which is part of an ongoing project, was to assess: 1) the validity and reliability of PPG-based HR wearable devices while exercising outdoors on a stationary bike, and 2) the effect of ambient light on HR values.
METHODS:
Twelve physically active males volunteered (age: 23.7±5.5 years; mean±SD; skin tone:2-6). Following a familiarization session, they performed three trials on a Monark bike (LC6, Sweden) separated by 20 min. Trials 1 and 2, which were identical, required participants to perform three 5-min bouts at age-predicted 50%, 60% and 70% of maximum HR with 5 min rest in between each bout. In trial 3, they performed 5-min cycling at 50% and 70% of HRmax with 5 min rest while wearing a long-sleeved white jersey to cover the PPG devices. This practice reduced ambient light intensity by ~50% (95,853±23,407 lux on the skin, about 3 cm away from the wrist-worn device, and 47,992±17,187 lux over the jersey at the same body location). HR was assessed with only one unit per device brand which were: a chest-worn HR monitor (Polar H10, Finland), previously tested for validity and served as the criterion device, a wrist-worn (Whoop Life, USA) and an arm-worn HR monitor (Coros, USA). The placement of the devices was as per the manufacturer’s guidelines. Validity was assessed with absolute error, mean absolute percentage error (MAPE), mean bias and 95% Limits of Agreement (LoA) between the tested devices and the chest-worn device. Reliability was assessed with the 95% prediction interval (PI) and ICC.
RESULTS:
Ambient temperature was 28.4±2.4 °C, and WBGT 19.3±2.0 °C. Depending on exercise intensity, absolute error and MAPE ranged between 1.0 bpm and 1.2 bpm and 0.8% to 1.2% for the wrist-worn device and between 1.2 bpm and 1.9 bpm and 1.0% and 1.9% for the arm-worn, respectively. The wrist-worn device showed small mean biases (range: −0.5 bpm to 0.2 bpm) and relatively narrow LoA (range: −4.1 bpm to 3 bpm), whereas the values for the arm-worn ranged from −1.1 bpm to 0.5 bpm (bias) and −7.5 bpm to 5.6 bpm (LoA). No difference in HR was observed between the tested and the criterion devices in the jersey trial. ICCs were relatively high (0.95-0.96) and 95% CI ranged between -9.2 bpm and +9.0 bpm for the chest-worn, -8.8 bpm and +8.6 bpm for the wrist-worn and -10.1 bpm and +10.5 bpm for the arm-worn device.
CONCLUSION:
Acknowledging the limitation of the small sample size, the preliminary findings of this study seem to indicate high accuracy of the PPG devices. Reliability varied between devices. Under the conditions of the present study, ambient light intensity did not appear to affect heart rate measurements.
(1)Muhlen et al. Br. J Sports Med 2021;55:767