Machine learning model uses data from wearable devices to detect COVID-19 in HCWs

Editor's Note

Researchers at the Icahn School of Medicine, Mount Sinai, New York City, have developed a machine learning model that uses physiological metrics collected from wearable devices that can detect and predict COVID-19 in healthcare workers (HCWs).

A total of 407 HCWs from 7 hospitals were enrolled in the study. Participants downloaded a custom smart phone app, wore an Apple Watch, and completed a daily questionnaire on how they felt and whether they had been diagnosed with COVID-19.

The researchers examined 5 machine learning approaches to determine which performed best to predict positive COVID-19 nasal PCR results.

They found that the machine learning algorithm called “gradient-boosting machines” (GBM) had the most favorable validation performance, with an average sensitivity of 82% and specificity of 77%.

The most important markers for predicting COVID-19 infection were heart rate variability, or the calculation of the small-time differences between each heartbeat..

Though further validation is necessary, this modality may be helpful to monitor large numbers of people for COVID-19 infection and help direct testing toward high-risk individuals, the researchers say.