The authors were interested in whether there is a correlation between certain laboratory test results and a COVID-19 diagnosis. They trained a machine learning model to analyse the correlation between SARS-CoV-2 test results and 20 routine laboratory tests collected within a 2-day period of the virology test. They examined on a cohort of 75,991 inpatients and outpatients who were tested for SARS-CoV-2 between March and July. 7335 of this cohort were positive by RT-PCR or antigen testing and had at least 15 of 20 lab results performed within the window period. Using these results, the authors’ model could predict the SARS-CoV-2 test result with a specificity of 86.8%, a sensitivity of 82.4% and an overall accuracy of 86.4%. Therefore, while molecular-based and antibody tests remain the reference standard method for diagnosing COVID-19, a complementary method could be used based on a fully independent set of indicators.
Read more here