Fully Automated Photoplethysmography-Based Wearable Atrial Fibrillation Screening in a Hospital Setting

Figures

• 

  Figure 1. Wristband with integrated PPG sensor.

• 

  Figure 2. Data flowchart. Data from the wearable device were continuously collected and transmitted via Bluetooth to a local gateway. Each wearable device was paired with a dedicated gateway (1:1). The gateway then forwarded the data securely over WLAN to a server, deployed on-site at the University Hospital Basel (UHBS). The server (Device Hub) maintained communication with multiple gateways (1:n) to monitor data flow, signal quality, and battery status. The deidentified raw PPG data were then sent from the server to an external server of Preventicus for analysis, where they were evaluated for conditions such as atrial fibrillation (AF), sinus rhythm (SR), or noise. An ECG-comparable PDF report was generated based on the results. In cases of suspected arrhythmia, the data were forwarded to the Telecare Center Ulm for further quality control and confirmation by trained technicians. The results of the analysis, along with the status of the quality control, were continuously synchronized with the server at the University Hospital Basel and re-identified with the patient ID for tracking and clinical reference.

• 

  Figure 3. Flowchart of the study inclusions, exclusions and full analysis set. AA, absolute arrhythmia; ECG, electrocardiography; PPG, photoplethysmography.

• 

  Figure 4. Time to first detection of AA in the PPG data. The black line indicates the Kaplan–Meier estimated cumulative incidence rate of first AA detection. The dotted black lines show the corresponding 95% confidence interval. The vertical dotted blue line indicates the 24 h mark. The total AA burden per hour from all patients with detected AA was examined (Figure 5). The AA burden per hour was 1.3 times higher during the night (22:00–06:00) compared to the day (p = 0.03) (corrected for better signal quality at night). The highest AA burden occurred from 03:00 to 03:59 in the night, followed by between 22:00 and 22:59. The lowest AA burden was observed from 16:00 to 16:59 in the afternoon.

• 

  Figure 5. Total AA burden per hour. Each bar represents the total time spent in AA within one hour from all patients with AA. For example, at hour 6, lasting from 06:00 to 06:59, 68 min of AA was identified. The bright blue bars are time intervals during the day and dark blue bars are time intervals during the night (22:00–06:00). The two hours with the greatest AA burden occurred at night.