The standard deviation of the inter-beats interval between QRS complexes recorded during 24 h (SDNN24) is considered the gold standard of Heart rate variability (HRV) features for cardiac health [1]. SDNN24 is an HRV feature that requires 24 h of continuous recording Inter-Beat Intervals, traditionally achieved using a Holter device, that makes the data collection difficult during people’s everyday life, therefore not performed routinely. Thanks to the technological advancements of recent decades, it is now possible and affordable to continuously record heart beats during 24 h via wrist-worn wearable devices equipped with heart rate sensors [2]. 

The low cost of these devices and their unobtrusiveness allows the larger part of the population to continuously and passively measure their heart activity. Wrist-worn wearable devices equipped with heart rate sensors have a great potential impact on the preventative health field, because it is now possible to estimate the users’ health status, capturing early signs of cardiac health deterioration [3]. Due to the fact that instruments able to record inter-beats intervals, such as wrist-worn wearable devices, are more comfortable to be worn by people during daily life compared to medical devices (e.g., Holter). They could, in theory, be used to estimate SDNN24. However, inter-beats intervals recorded from these devices suffer from high amounts of noise and motion artifacts that propagate to HRV features [4,5]. Furthermore, wrist-worn devices normally only report heart rate data (HR), as these have been proven to be more reliable than inter-beats intervals data which can only be estimated from these devices in both resting and during physical activity with a small error [3-6], that seldom report the inter-beats intervals data needed to compute SDNN24.

The study of Morelli et al. [7] has shown that SDNN24 may possibly be estimated from HR data, without the need to have 24-hour inter-beat intervals data available. Continuous or semicontinuous HR measures are nowadays affordable via wearable devices equipped with a PPG sensor, such as fitness trackers. Estimating HRV features from wearable devices is known to be problematic because of noise induced by motion artefacts that affect the frequencies normally used to assess the activity of the Autonomic Nervous System, such as the Root Mean Square of Successive Differences (RMSSD) and the Sympathovagal Imbalance (SVI) [4,5]. However, HR is calculated by taking the average of the duration of the inter-beat intervals’ time-series that is equivalent to applying a low pass filter that permits one to filter out most of the noise. This makes the SDNN24 estimation particularly resilient to noisy data (i.e., unevenly sampled data with huge quantities of missing data induced by motion artefacts) a condition normally found with wrist-worn wearable devices. Finally, this result indicates that HR fitness trackers have the potential to implement the continuous monitoring of cardiovascular health from passively collected data that could enable targeted interventions at early signs of deterioration of Sinoatrial Node activity. 

Written by: Alessio Rossi

SoBigData++ micro-project: SDNN24 estimation from semi-continuous HR measures

Reference

[1] Shaffer, F.; Ginsberg, J. An Overview of Heart Rate Variability Metrics and Norms. Front. Public Health 2017, 5, 258, doi:10.3389/fpubh.2017.00258.

[2] Haghi, M.; Thurow, K.; Stoll, R. Wearable Devices in Medical Internet of Things: Scientific Research and Commercially Available Devices. Healthc. Inform. Res. 2017, 59, 4–15.

[3] Li, K.H.C.; White, F.A.; Tipoe, T.; Liu, T.; Wong, M.C.; Jesuthasan, A.; Baranchuk, A.; Tse, G.; Yan, B.P. The current state of mobile phone apps for monitoring heart rate, heart rate variability, and atrial fibrillation: Narrative review. JMIR mHealth and uHealth 2019, 7, e11606, doi:10.2196/11606.

[4] Morelli, D.; Rossi, A.; Cairo, M.; Clifton, D. Analysis of the Impact of Interpolation Methods of Missing RR-Intervals Caused by Motion Artifacts on HRV Features Estimations. Sensors 2019, 19, 3163, doi:10.3390/s19143163.

[5] Morelli, D.; Bartoloni, L.; Colombo, M.; Clifton, D. Profiling the propagation of error from PPG to HRV features in a wearable physiological-monitoring device. Healthc. Technol. Lett. 2018, 5, 59–64,

[6] Jo, E.; Lewis, K.; Directo, D.; Kim, M.; Dolezal, B. Validation of BiofeedbackWearables for Photoplethysmographic Heart Rate Tracking. J. Sport Med. 2016, 15, 540.

[7] Morelli, D.; Rossi, A.; Bartoloni, L.; Cairo, M.; Clifton, D.A. SDNN24 estimation from semi-continuous HR measures. Sensors 2021, 21, 1463. https://doi.org/10.3390/s21041463