DETERMINING PULSE-RATE FROM WRIST PLACED ACCELEROMETRY IN CHILDREN IN ORDER TO IMPROVE ESTIMATES OF SLEEP.

Authors

Anna K. Maxwell, University of South CarolinaFollow
Robert G. Weaver, University of South CarolinaFollow
Bridget Armstrong, University of South CarolinaFollow
James White, University of South CarolinaFollow
Olivia Finnegan, University of South CarolinaFollow
Aliye Cepni, University of South CarolinaFollow
Zhu Xuanxuan, University of South CarolinaFollow
Lauren Von Klinggraeff, University of South CarolinaFollow
Hannah Parker, University of South CarolinaFollow
Meghan Bastyr, University of South CarolinaFollow
Nicholas Niako, University of South CarolinaFollow
Elizabeth Adams, University of South CarolinaFollow
Sarah Burkart, University of South CarolinaFollow
Michael Beets, University of South CarolinaFollow
MVS ChandrashekharFollow

Abstract

BACKGROUND: Children’s free-living sleep is most commonly measured via wrist-placed accelerometry. Similar to taking someone’s pulse, wrist-placed accelerometry may be sensitive enough to detect pulse-rate. This may be critical because current sleep detection methods using accelerometry rely on movement alone to detect sleep. This leads to poor detection of wake and the inability to detect sleep stages. Adding a physiological signal like pulse rate to sleep detection addresses both these limitations. The objective of this study was to estimate children’s pulse-rate via wrist placed accelerometry and compare these estimates to electrocardiogram (ECG) as a gold standard. METHODS: Participating children wore a consumer wearable (Apple Watch Series 7) and a wrist-placed research grade accelerometer (Actigraph GT9X) while undergoing an overnight laboratory-based polysomnography (PSG), including a 3-lead ECG. Raw accelerometry data from the Apple device was extracted using SensorLog, a freely available user-written application that leverages the devices’ application programming interface. Actigraph data was extracted via Actilife Software. All subsequent processing was performed in MATLAB. Pulse-rate estimates from the raw accelerometry data were calculated from peak magnitude frequency in short time Fourier Transforms of Hilbert transformed jerk computed from acceleration magnitude. Heartrates from the criterion ECG were estimated from R-R spacings using R-pulse detection in normalized ECG traces. Mean absolute error (MAE) and mean absolute percent error (MAPE) were calculated to assess agreement between the accelerometry estimated pulse-rate from Actigraph and Apple and the ECG estimated heartrate. RESULTS: Eighty-four 5-12-year-old children (63% male, 72% White, 66% with mild/moderate obstructive sleep apnea) participated. One child was excluded as the ECG data stream was corrupted during collection. For Actigraph MAE and MAPE were high at 39(SD=20) beats/minute and 49.0%(SD=27.4%). For Apple MAE and MAPE were much lower at 8.9(SD=6.2) beats/minute and 10.2%(SD=6.5%) CONCLUSIONS: Raw accelerometry data extracted from Apple but not Actigraph can be used to estimate pulse-rate in children while they sleep. Future work is needed to explore the sources (i.e., hardware, software, etc.) of Actigraph’s relatively poor performance.

Recommended Citation

Maxwell, Anna K.; Weaver, Robert G.; Armstrong, Bridget; White, James; Finnegan, Olivia; Cepni, Aliye; Xuanxuan, Zhu; Von Klinggraeff, Lauren; Parker, Hannah; Bastyr, Meghan; Niako, Nicholas; Adams, Elizabeth; Burkart, Sarah; Beets, Michael; and Chandrashekhar, MVS (2024) "DETERMINING PULSE-RATE FROM WRIST PLACED ACCELEROMETRY IN CHILDREN IN ORDER TO IMPROVE ESTIMATES OF SLEEP.," International Journal of Exercise Science: Conference Proceedings: Vol. 16: Iss. 3, Article 81.
Available at: https://digitalcommons.wku.edu/ijesab/vol16/iss3/81