RESUMO
Severity prediction of COVID-19 remains one of the major clinical challenges for the ongoing pandemic. Here, we have recruited a 144 COVID-19 patient cohort consisting of training, validation, and internal test sets, longitudinally recorded 124 routine clinical and laboratory parameters, and built a machine learning model to predict the disease progression based on measurements from the first 12 days since the disease onset when no patient became severe. A panel of 11 routine clinical factors, including oxygenation index, basophil counts, aspartate aminotransferase, gender, magnesium, gamma glutamyl transpeptidase, platelet counts, activated partial thromboplastin time, oxygen saturation, body temperature and days after symptom onset, constructed a classifier for COVID-19 severity prediction, achieving accuracy of over 94%. Validation of the model in an independent cohort containing 25 patients achieved accuracy of 80%. The overall sensitivity, specificity, PPV and NPV were 0.70, 0.99, 0.93 and 0.93, respectively. Our model captured predictive dynamics of LDH and CK while their levels were in the normal range. This study presents a practical model for timely severity prediction and surveillance for COVID-19, which is freely available at webserver https://guomics.shinyapps.io/covidAI/.
