Pre-symptomatic detection of COVID-19 from smartwatch data

The COVID-19 pandemic has disrupted societies and scientific research around the globe, and necessitates rapid innovation among the microbiology community to address the unprecedented viral crisis. Critical to combatting spread of the SARS-CoV-2 virus is understanding the factors that predict infection and hospitalization in personalized ways to allow unique responses. Leveraging individual level data generated from wearable technology, we have been developing predictive systems for COVID-19 using physiological measures from smartwatches and other wearable devices to enable a real-time pre-symptomatic COVID-19 alarm system through our phone application (https://innovations.stanford.edu/wearables).

Tens of millions of Americans actively use wearable devices digitally measuring vital signs on a daily basis, and these have proven useful for monitoring health and illness onset with potential for real-time monitoring and disease detection. Using smartwatch data from infected individuals in a cohort of over 6,000 participants, we investigated the use of wearables for early, pre-symptomatic detection of COVID-19. In phase 1, we demonstrated that COVID-19 infections are associated with alterations in heart rate, steps and sleep in 80% of COVID-19 infection cases, with detection before or at symptom onset in 85% of the positive cases. We developed a method detecting 67% of infection cases at or before symptom onset in real-time, which has been implemented in phase 2 with currently over 3,000 participants receiving alerts from our custom phone application to warn of potential COVID-19 infection signs. Results are being validated in the second phase through FDA certified nasal swabs, as well as blood and stool microsampling kits mailed to study participants. Stool samples will be used for microbiome analyses and compared to extensive survey data, wearable data, and predictive algorithm accuracy.