Abstract
To evaluate the accuracy of the smartwatch in estimating carotid-femoral pulse wave velocity (cfPWV). A cohort of gender-matched volunteers aged 18-80 years were recruited. At the sitting and supine positions, cfPWV was measured alternately by smartwatch and CompliorAnalyse, for each participant, and nine sets of data were collected from each participant with a 60 s interval between measurements. The accuracy of cfPWV measurement for smartwatches was assessed using mean error (ME) and mean absolute error (MAE), while the consistency of the two methods was assessed using the Bland-Altman analysis and concordance class correlation. A total of 347 participants were enrolled. The mean cfPWV was 9.01 ± 2.29 m/s measured by CompliorAnalyse and 9.06 ± 1.94 m/s by smartwatch. The consistency correlation coefficient (CCC) was 0.9045 (95% CI 0.8853-0.9206), the ME was 0.046 ± 0.92, and the MAE was 0.66 (95% CI 0.59-0.73). Bland-Altman analysis showed that the error of 95% samples was in the range between -1.77 m/s and 1.86 m/s. The Kappa value of cfPWV greater than 10 m/s was 0.79, the area under the ROC curve was 0.97 (P < 0.001), sensitivity was 0.90, specificity was 0.93, positive predictive value was 0.83 and negative predictive value was 0.96. Smartwatch can accurately estimate cfPWV to evaluate arterial stiffness. This method is simple and feasible and is suitable for people to actively and early monitor vascular elasticity.