Poster: Heart Rate Performance of a Medical-Grade Data Streaming Wearable Device

Wrist-worn devices afford convenient and unobtrusive heart rate sensing, however, motion artifacts can lead to unreliable data recordings. This paper evaluates heart rate estimates acquired during treadmill walking and 12 hours of everyday living from a medical-grade Empatica E4 data streaming wristband wearable compared to a Polar H10 chest strap ECG sensor. For treadmill walking, heart rate Mean Absolute Percentage Errors (MAPEs) were between 7.2% and 29.2%, and IntraClass Correlations (ICCs) between 0.6 and -0.5, indicating moderate agreement and strong disagreement, respectively. During 12-hour everyday living acquisitions, heart rate estimate MAPEs were between 5.3% and 13.5% and ICCs between 0.7 and 0.1, indicating good to poor agreements.