COVID-19 diagnosis prediction in emergency care patients: a machine learning approach

Andre Filipe de Moraes Batista; Joao Luiz Miraglia; Thiago Henrique Rizzi Donato; Alexandre Dias Porto Chiavegatto Filho

During the last few weeks, the first studies were published that apply machine learning to predict different COVID-19 outcomes. 8 So far only one study analyzed diagnosis of COVID-19 with routinely-collected data, in this case to predict cases using as controls patients with viral pneumonia by applying logistic regression. 9 Another study of 53 patients from two hospitals in Wenzhou, China, analyzed the accuracy of five machine learning algorithms to predict Acute Respiratory Distress Syndrome (ARDS) in patients with COVID-19. 10 Two other studies applied machine learning algorithms to predict mortality in patients with COVID-19, using patient data from Kaggle and China. 11, 12 We propose that machine learning algorithms can be used to allocate priorities in receiving the RT-PCR tests in the case of a shortage, and also to help with critical care decisions while the RT-PCR results are being processed (which have been frequently taking more than a week in most places of Brazil). A promising area for future research will also be to analyze the combined performance of the new rapid tests and the machine learning algorithms.