Background
Monitoring of oxygen saturations (SpO2) in COVID-19 patients being managed at home has been recommended to detect hypoxia. Recent advances in smartwatch technology have provided consumers the ability to record blood oxygen levels, however their accuracy has not been well-established in comparison to standard photoplethysmography (PPG).
Aim
To assess the accuracy of smartwatch (Apple Watch and Withings ScanWatch) derived SpO2 readings compared to standard ward-based PPG.
Methods
Patients diagnosed with COVID-19 were prospectively recruited between November 2021 and February 2022. Simultaneous smartwatch and ward PPG readings were obtained from the left and right wrists consecutively. For patients on supplemental oxygen with an SpO2 ≥95%, repeat recordings were obtained 2 minutes following oxygen removal. The primary endpoint was smartwatch detection of hypoxia (SpO2 ≤92%) compared to gold standard ward-based PPG monitoring.
Results
A total of 526 smartwatch readings were obtained from 123 patients (male 53.7%, age 63±19 years). For detection of hypoxia, compared to ward-based PPG, the Apple Watch had a sensitivity and specificity of 30.5% and 92.9% respectively, with a positive predictive value (PPV) of 64.1% and negative predictive value (NPV) of 76.2%. The Withings ScanWatch had a sensitivity and specificity of 62.2% and 69.9% respectively with a PPV 46.4% and NPV 81.6%. Overall accuracy was 74.5% for the Apple Watch and 67.6% for the Withings watch.
Conclusion
While smartwatch technology is able to provide SpO2 readings, its accuracy does not appear to be sufficient to replace standard PPG technology in monitoring hypoxia in home-based COVID-19 patients.
| Sensitivity | Specificity | PPV | NPV | PLR | NLR | Accuracy | |
|---|---|---|---|---|---|---|---|
| Apple | 30.5% | 92.9% | 64.1% | 76.2% | 4.27 | 0.75 | 74.5% |
| Withings | 62.2% | 69.9% | 46.4% | 81.6% | 2.07 | 0.54 | 67.6% |
534