Digital SARS-CoV-2 Detection Among Hospital Employees: Participatory Surveillance Study

Abstract

Background The implementation of novel techniques as a complement to traditional disease surveillance systems represents an additional opportunity for rapid analysis. Objective The objective of this work is to describe a web-based participatory surveillance strategy among health care workers (HCWs) in two Swiss hospitals during the first wave of COVID-19. Methods A prospective cohort of HCWs was recruited in March 2020 at the Cantonal Hospital of St. Gallen and the Eastern Switzerland Children’s Hospital. For data analysis, we used a combination of the following techniques: locally estimated scatterplot smoothing (LOESS) regression, Spearman correlation, anomaly detection, and random forest. Results From March 23 to August 23, 2020, a total of 127,684 SMS text messages were sent, generating 90,414 valid reports among 1004 participants, achieving a weekly average of 4.5 (SD 1.9) reports per user. The symptom showing the strongest correlation with a positive polymerase chain reaction test result was loss of taste. Symptoms like red eyes or a runny nose were negatively associated with a positive test. The area under the receiver operating characteristic curve showed favorable performance of the classification tree, with an accuracy of 88% for the training data and 89% for the test data. Nevertheless, while the prediction matrix showed good specificity (80.0%), sensitivity was low (10.6%). Conclusions Loss of taste was the symptom that was most aligned with COVID-19 activity at the population level. At the individual level—using machine learning–based random forest classification—reporting loss of taste and limb/muscle pain as well as the absence of runny nose and red eyes were the best predictors of COVID-19.