Real-time alerting systems for infections using wearable data leverage advanced wearable technology to detect early signs of COVID-19 and influenza in solid organ transplant recipients (SOTR). Utilizing smartwatches, the system collects physiological data, including heart rate, step count, and sleep patterns, analyzed in real-time to identify deviations from the individual's baseline, indicative of infection. 3,318 participants were enrolled, with data collected via smartwatches. The MyPHD app facilitated participant enrollment and data collection, ensuring secure data transfer. The system employed three detection algorithms: NightSignal, RHRAD, and CuSum. NightSignal, the primary algorithm, provided real-time alerts by detecting outlier signals that could signify early signs of infection. Participants received and were required to annotate daily alerts through surveys on symptoms, activities, and COVID-19 test results. The alerts were presented as either red or green, indicating abnormal or normal physiological states, respectively. Among the participants, 84 tested positive for SARS-CoV-2, with 80% of infections being detected at or before the onset of symptoms. The median lead time for pre-symptomatic detection was approximately three days. NightSignal demonstrated a high sensitivity (80%) for detecting COVID-19 infections pre-symptomatically. Additionally, other stress events, though triggering fewer alerts, were detectable. Preliminary results indicate the wearable-based monitoring system is effective in early detection of infections among SOTR. Future research will focus on refining the algorithms, particularly enhancing the specificity and reducing false-positive rates. Expanding the system's application to a broader population and other infectious diseases could further demonstrate its scalability and utility in public health.
