Accuracy of photoplethysmography-based measurement of blood pressure

Abstract

Abstract Funding Acknowledgements Type of funding sources: None. Introduction Photoplethysmography (PPG)-based blood pressure (BP) measurements are available in the latest smartwatches and are thus available for patients in the consumer market. While the devices typically comply with the medical device regulation, accuracy is often not known and data on validation are not publicly available. Purpose To investigate the accuracy and precision of the Samsung Galaxy Watch Active 2 PPG-based BP measurement as compared to conventional BP measurement and to investigate clinical applicability. Methods 31 patients scheduled for ambulatory 24-hour BP measurement were included in the study. 24-hour BP devices were attached to the non-dominant arm, smartwatches were attached to the opposite arm. Calibration was performed as instructed by the manufacturer. The smartwatch was used to register BP intermittently for 24 hours. Results from both devices were compared using the Taffé method. Results and discussion Mean age was 59 years. 17 patients (55%) were male. Comparison of the two methods is demonstrated in Figure 1. The smartwatch overestimates the systolic BP up to approximately 132 mm Hg and then underestimates the systolic BP, illustrating the presence of proportional (0.48) and differential (69.0) bias. Precision for the gold standard method was higher compared to the smartwatch measurements. For increasing values of systolic BP, the precision of the smartwatch decreases. For diastolic BP, differential (31.4) and proportional bias (0.68) are present, resulting in an overestimation of the diastolic BP at lower values (<100 mmHg). Precision for the gold standard method was higher at lower BP values, while precision for smartwatch measurements was higher at higher BP values (i.e. there was a lower standard deviation of measurement error). Figure 1 Bias and precision plots for systolic and diastolic blood pressure (BP). In each figure the x-axis indicates the best linear unbiased prediction (BLUP) or true value of BP. Panel A and B indicate the bias plots. Smartwatch measurements (blue dashed lines) are compared to reference measurements (solid black line); both values are read from the leftmost y-axis. The bias (difference of smartwatch measurement compared to the reference method) is depicted as a solid red line; the value of bias is read from the rightmost y-axis. Panel C and D indicate the precision plots, indicating the standard deviation of measurement errors for both measurement methods. The reference method is depicted in black; the smartwatch method is depicted in blue. Conclusion There is a differential and proportional bias in PPG-based BP measurement of systolic and diastolic BP measurement. The systolic BP is overestimated at lower values and underestimated at higher values. This might indicate potential for application as a screening tool for hypertension. Further research is needed to investigate clinical applicability.