Multi-Modal Wireless COVID Monitoring & Infection Alerts for Concentrated Populations
Abstract: The high aerosolized transmissibility of COVID, long asymptomatic incubation period,
and highly variable presentation attributes of the COVID pandemic have proven challenging in
many settings where patchwork pandemic responses have disproportionately negatively
impacted vulnerable socioeconomic, minority, and disabled sub-populations. Unfortunately, these
dire trends are only made more acute in settings that feature populations with limited mobility and
little to no ability to self-isolate (dense concentrated populations [DCPs]), such as residential
nursing homes, schools, drug rehabilitation services, prison and psychiatric facility populations,
and high-frequency essential medical services, such as chemotherapy infusion clinics or dialysis
units. In these DCP settings, limited diagnostic testing, prolonged indoor contact, limitations in
cleaning and filtration capacities, support staff shortages, pre-existing comorbidities, and lack of
effective infectious disease surveillance systems all collude to drive an increased COVID burden
in DCPs. From this, it is clear that alternative detection strategies for DCPs are urgently needed
to improve local capacity to monitor COVID outbreaks, mitigate their spread, and thus reduce
inequitable disease and mortality burdens in these under-resourced and often overcrowded
settings. In previous work, we developed a first generation detection system using heart rate data
from commercially-available Fitbit Ionic wearable devices to detect the onset of COVID and other
infectious diseases up to 10 days before users self-reported symptom onset (overall sensitivity
67% prior to symptom onset). Here, we propose to further develop this system for the improved
detection of COVID and other infectious diseases in DCPs using existing wearable fitness devices
in a wireless and interoperable digital health framework that centralizes all wearable-derived data
on PHD while tailoring its presentation and health event alert system to the IT capabilities and
needs of each DCP setting. In this, not only will we adapt our existing infection detection
algorithms for each DCP’s particular baseline characteristics, IT infrastructure, and needs, but
also use incoming data to further optimize the performance of those algorithms for continuous
improvement in the sensitivity, specificity, and alert lead time for COVID onset. This will quickly
enable under-resourced DCP support staff to access and use world-class COVID surveillance
data in identifying individual infection events, implementing isolation, cleaning, and testing
policies, and minimizing transmission, thus reducing the burden of COVID in DCP settings and
reducing DCP morbidity and mortality overall.