Real-time alerting system for COVID-19 and other stress events using wearable data.
Level 3 - non-randomized controlled study
Prospective cohort study evaluating a diagnostic/alerting monitoring algorithm
PubMed 34845389 · doi:10.1038/s41591-021-01593-2
What was done
Researchers developed and prospectively tested a real-time alerting system that continuously analyzes smartwatch physiological and activity data (heart rate and step count) to detect early aberrant signals of infection. The platform was evaluated in a prospective cohort of 3,318 participants, tracking alerts alongside participant survey responses regarding symptoms, COVID-19 infection, and other daily stressors.
What was found
Among 84 participants infected with SARS-CoV-2, the alerting system generated pre-symptomatic or asymptomatic alerts in 67 individuals (80%). Pre-symptomatic signals occurred at a median of 3 days prior to the onset of symptoms. Non-COVID events (including other respiratory infections, psychological stress, alcohol intake, and travel) also generated alerts, but at a lower mean frequency (1.15 alert days per person) compared to confirmed COVID-19 cases (3.42 alert days per person).
Why it matters
This demonstrates the feasibility of utilizing consumer wearable devices for automated, real-time early warning of viral infections days before clinical symptoms appear. If deployed at scale, such algorithms could prompt earlier self-isolation and diagnostic testing.
Limits
The alerting mechanism is non-specific, triggering alerts due to ordinary physiological stressors such as alcohol consumption, travel, and emotional stress. The total number of confirmed COVID-19 cases in the cohort was modest (84 individuals), and the abstract does not report overall sensitivity, specificity, or wear-time compliance across the entire cohort.
Cited by
- supports A real-time wearable monitoring algorithm detected COVID-19 and alerted individuals at or before symptom onset in 44 out of 63 cases (70%).