Accuracy of wearable smartwatch for measuring blood pressure and oxygen saturation across a wide altitudinal gradient: a comparative study in Migrant and Resident populations

TL;DR

This study tested a smartwatch's accuracy for measuring blood pressure and oxygen saturation at different altitudes and found it reliable for oxygen but less so for blood pressure.

Contribution

The study evaluates a commercial smartwatch's performance across altitudes and assesses the impact of adaptation and ethnicity on accuracy.

Findings

SpO2 accuracy remained within ISO thresholds across all altitudes tested.

Blood pressure measurements showed significant deviations from ISO standards at higher altitudes.

Measurement error variability was higher in acutely exposed individuals.

Abstract

Wearable devices are increasingly used to monitor physiological parameters, yet their accuracy at high altitude remains uncertain. We evaluated a commercial smartwatch for peripheral oxygen saturation (SpO2) and blood pressure (BP) measurement across a wide altitudinal gradient and examined whether adaptation status and ethnicity influence accuracy.

A prospective observational study was conducted at four sites in Western China (500–4,014 m). The Migrant cohort (n = 24) comprised lowlanders assessed longitudinally at all sites; the Resident cohort (n = 85) comprised long-term high-altitude residents sampled cross-sectionally. Smartwatch measurements (HUAWEI Watch GT3 Pro) were compared with reference devices using Bland–Altman analysis and benchmarked against ANSI/AAMI/ISO standards.

SpO2 accuracy was maintained across altitudes, with RMSE of 0.19%–0.81%, within the ISO 80601-2-61:2017 threshold (≤3%). BP showed wider limits of agreement with a persistent negative bias. SBP in Migrants at 2,560 m exceeded ISO 81060-2:2018 criteria for both bias (−5.29 mmHg) and SD (9.88 mmHg; threshold ≤8 mmHg). DBP bias in Migrants at 4,014 m also exceeded the ±5 mmHg threshold (−5.28 mmHg). No significant differences in measurement error were found between Migrants and Residents or between Han and Zang ethnicities after FDR correction, though error variability was markedly higher in acutely exposed individuals (variance ratio up to 17.5 for SpO2).

Smartwatch SpO2 readings remained within ISO benchmark thresholds, supporting trend monitoring utility. BP readings showed specific deviations from ISO criteria during acute exposure and warrant cautious interpretation above 2,500 m.

Figures