Movement, Sweating, and Contact Pressure as Sources of Heart Rate Inaccuracy in Wearable Devices

Wearable devices enable continuous heart rate (HR) monitoring through photoplethysmography (PPG). The impact of wrist-worn devices' sensor contact pressure and sweat, and of their interaction with movement, on HR monitoring is unclear. HR was recorded in 17 healthy individuals using two smartwatches, Garmin Vivoactive 4 (GV) and Fitbit Sense (FS), concurrently with ECG at rest and during controlled arm movement at three increasing intensities. Recordings were repeated after reducing contact pressure by loosening the wristband by one or two notches and using one or two drops of saline solution to simulate sweating. In optimal conditions, the mean absolute percentage error (MAPE) was (median [interquartile range]) 4.3% (1.4%, 7.7%) and 3.1% (1.6%, 5.0%)
$(p=0.58)$
, for GV and FS, respectively. Loosening the wristband by 1 notch increased MAPE for FS during rest
$(p=0.021)$
, moderate
$(p=0.004)$
and vigorous
$(p=0.002)$
movement, but not for GV, for which loosening the wristband by 2 notches increased MAPE during moderate
$(p=0.015)$
and vigorous
$(p=0.008)$
movement. Simulated sweat increased MAPE during moderate movement using
$FS(p{=0.002})$
, and during vigorous movement for both devices. In conclusion, contact pressure and sweating can increase HR inaccuracy even during rest and moderate movement.