Disclaimer: Due to health regulations and clearances, some ScanWatch features may not be available or may not have clinical validation in your region. Learn more.
Study on ECG Accuracy and Atrial Fibrillation Detection Algorithm
Atrial Fibrillation Detection Accuracy Study (via the Photoplethysmography Optical Sensor)
Study of the Accuracy of SpO2 Measurements
Study on ECG Accuracy and Atrial Fibrillation Detection Algorithm
A clinical study (ClinicalTrials.gov Identifier: ongoing) was conducted to validate that ScanWatch could perform a medical-grade electrocardiogram and detect atrial fibrillation with these tracings. ScanWatch ECG recordings were compared to 12-lead ECG recordings and reviewed in a blind assessment by 3 cardiologists. The study was approved by an ethics committee. It was carried out according to Good Clinical Practice, the ISO 14155 standard, and the Declaration of Helsinki. Each participating patient gave written consent.
The study took place in several centers, including Centre de Cardiologie du Nord (CCN) & Hôpital Européen George Pompidou (HEGP), using a sample of varied age, height, weight and sex. Approximately half of participants had atrial fibrillation. Physicians simultaneously recorded an electrocardiogram on ScanWatch and a 12-lead baseline electrocardiogram on each patient before comparing the tracings. They used the reference EKG to make a diagnosis. They also conducted a blind review of the ECGs from the reference ECG or the watch to analyze the quality of the waveforms and waveforms (P wave, QRS wave, T wave, and duration of PR/QT/QRS intervals).
The results showed that:
- ScanWatch measures a trace that is extremely consistent with the reference on all the characteristic waves of an ECG.
- The atrial fibrillation (Afib) detection algorithm demonstrated a sensitivity and specificity of 1 (lower bound of the confidence interval at .92 and .94 respectively) on record for the classification of ECGs as AF or normal sinus rhythm. One-fifth of the records were classified as inconclusive by the algorithm because they showed poor signal quality or arrhythmias of other types than Afib.
Atrial Fibrillation Detection Accuracy Study (via the Photoplethysmography Optical Sensor)
In the previous study, another measurement was taken on the same patients with a photoplethysmography (optical) sensor to detect atrial fibrillation. This sensor is less accurate, but the results were nevertheless excellent: the sensitivity/specificity of detection of atrial fibrillation were .98 and .93 (lower bound of the confidence interval at .90 and .84 respectively). One-fifth of the records were classified as inconclusive by the algorithm in this case as well.
Study of the Accuracy of SpO2 Measurements
A clinical study (ClinicalTrials.gov Identifier: NCT04380389) was conducted to validate that ScanWatch could deliver a medical-grade blood oxygen saturation measurement. The ScanWatch measurements were compared to those measured by gas analyses of the participants' blood samples. The study was approved by an ethics committee. It was conducted in accordance with Good Clinical Practice, ISO 14155, and the Declaration of Helsinki. Each participating patient gave written consent.
The study took place at the Hypoxia Lab of the University of California, San Francisco, according to ISO 80601-2-61. It was conducted in the laboratory on healthy subjects in hypoxia, with saturation controlled in steps of 70% to 100%, allowing us to evaluate the accuracy of our sensor over the entire useful range of SpO2.
The accuracy of ScanWatch, evaluated according to this ISO standard, is 3.0% RMSE. The FDA requires less than 3.5% in "Pulse Oximeters - Premarket Notification Submissions: Guidance for Industry and Food and Drug Administration Staff" and ISO 80601-2-61:2017 requires less than 4.0% for reflective oximeters.