Wrist-based wearable oscillometric blood pressure monitoring during Ramadan: a comparative study with standard devices

The determination of blood pressure (BP) is crucial for the regulation of cardiovascular health status, particularly under the condition of physiological stress as seen during Ramadan fasting. Conventional BP measurement techniques are, accurate, intermittent and do not reflect the dynamic changes of fasting and rehydration. Wrist-based wearable oscillometric blood pressure monitoring devices, incorporating miniaturized inflatable cuffs, provide a portable and non-invasive approach for assessing BP outside traditional clinical settings. To assess the reliability of a consumer-grade wearable device (Huawei Watch D) for measuring systolic and diastolic blood pressure during Ramadan fasting, and to compare its performance with a validated electronic upper-arm monitor after pre-fast and post-fast conditions (Sahoor and If tar). Methods: A within subject, repeated-measures investigation was performed with 101 healthy fasting subjects (ages 22-70). Each participant recorded BP twice daily during Sahoor and If tar using both the Huawei Watch D and an upper-arm electronic monitor, both are FDA-approved. Statistical methods comprised paired t tests, as well as Pearson correlations and Bland-Altman plots. To evaluate prediction accuracy and discover subgroups, we also used machine learning approaches (linear regression and K-means clustering). Results: Compared with the electronic upper-arm monitor, the Huawei Watch D demonstrated minimal mean bias for both systolic and diastolic blood pressure (SBP: +0.9 to +1.4 mmHg; DBP: -0.6 to -0.2 mmHg). However, SBP exhibited wide limits of agreement (±18-21 mmHg), whereas DBP showed narrower dispersion (±11-13 mmHg). Machine-learning models achieved moderate prediction accuracy for DBP (MAE ≈ 4.3-6.1 mmHg, R² up to 0.74), while SBP prediction remained less accurate (MAE ≈ 6.5-8.9 mmHg, R² up to 0.64).However, SBP exhibited wide limits of agreement (±20 mmHg), limiting diagnostic interchangeability. DBP showed narrower error dispersion, supporting the use of wearable devices for trend monitoring and screening rather than clinical diagnosis. Conclusions: Smartwatch BP estimates are associated with standard devices, however, individual differences are considerable and therefore do not allow for diagnostic interchangeability. Wearables, such as the Huawei Watch D, provide the interesting data on the BP trend during fasting and rehydration, especially in diastolic pressure. These devices may assist with culturally appropriate monitoring and early identification of BP abnormalities however are in need of validation in a clinical context.