Accuracy and reliability of wristband smartwatch-based estimation of circadian rhythm of heart rate in stroke patients

Abstract

Background & Objective: Stroke has a high incidence, disability, and mortality rate, which has a serious impact on the patients’ quality of life. Heart rate (HR) is valuable in the prognostic assessment of stroke patients. Continuous HR tracking helps to detect abnormal indicators and provide interventions timely to avoid disease progression. However, the cuff-based or electrode-based HR monitoring tools commonly used in clinic are not suitable for continuous monitoring in daily life. The aim of this study was to verify the accuracy and reliability of a wristband smartwatch in monitoring HR over 24-hour in stroke patients, including in terms of the circadian rhythm characteristics of HR.

Methods: 25 stroke patients underwent 24-hour HR monitoring by wearing a smartwatch and a clinical monitor. We use ambulatory blood pressure monitors and bedside monitors as the gold standard for clinical HR monitoring. The degree of association between the two measurements was compared using Pearson’s correlation coefficient. Bland-Altman analysis showed the accuracy of the smartwatch in HR monitoring. Intraclass correlation coefficients (ICC) and coefficients of variation (CV) were used to verify the reliability of the wristband smartwatch for repeated measurements.

Results: A total of 625 heart rate pairs were acquired. The smartwatches had an average error of 0.816 bpm in 24-hour, with no difference from standard monitors (P>0.05), indicating good accuracy. The intraday repeated measures ICC was 0.969 and the interday repeated measures ICC was 0.788, indicating good reliability.

Conclusions: The smartwatch is acceptable in detecting 24-hour ambulatory HR in stroke patients, and in the future, long-term HR monitoring with diurnal variations can be performed in daily life.