Compared to expert evaluation of the images, the neural network achieved upwards of 99% specificity, showing promise for the automated detection of COVID-19 infection in clinical settings.