An algorithm that is fed data from wearable fitness trackers can alert people of COVID-19 infection or stress before the onset of symptoms, according to a study published Nov. 29 in Nature.
The study followed over 3,000 participants with smart watches that measure physiological and activity signals such as heart rates and step counts. The health information was synced through an app, sent to the cloud and analyzed in real time.
The analysis of the data used an algorithm that can use detected physiological changes to alert users of new infections, including COVID-19. If the system detected a change, it would send the participant a real-time alert about a difference, annotating the alert as red. It also sent alerts when no change was detected. A green alert signaled everything was normal. Participants were then asked to supplement the alert with a survey of their symptoms.
The researchers found that the system could detect COVID-19 on or before the onset of symptoms in 80 percent of cases where a participant tested positive. It also gave alerts for asymptomatic cases, although the number of true asymptomatic cases is unknown due to a lack of testing. The median alert time was three days before the onset of symptoms.
Many of the other alerts given to participants were attributable to abnormal events, such as excessive alcohol consumption, stress or poor sleep. Seventy-three percent of participants found the frequency of the alerts acceptable, suggesting a potential commercial future.