Background: The outbreak of coronavirus disease 2019 (COVID-19) has globally strained medical resources and caused significant mortality.Objective: To develop and validate a machine-learning model based on clinical features for severity risk assessment and triage for COVID-19 patients at hospital admission.Method: 725 patients were used to train and validate the model. This included a retrospective cohort from Wuhan, China of 299 hospitalised COVID-19 patients from 23 December 2019 to 13 February 2020, and five cohorts with 426 patients from eight centres in China, Italy and Belgium from 20 February 2020 to 21 March 2020. The main outcome was the onset of severe or critical illness during hospitalisation. Model performances were quantified using the area under the receiver operating characteristic curve (AUC) and metrics derived from the confusion matrix.Results: In the retrospective cohort, the median age was 50 years and 137 (45.8%) were male. In the five test cohorts, the median age was 62 years and 236 (55.4%) were male. The model was prospectively validated on five cohorts yielding AUCs ranging from 0.84 to 0.93, with accuracies ranging from 74.4% to 87.5%, sensitivities ranging from 75.0% to 96.9%, and specificities ranging from 55.0% to 88.0%, most of which performed better than the pneumonia severity index. The cut-off values of the low-, medium- and high-risk probabilities were 0.21 and 0.80. The online calculators can be found atwww.covid19risk.ai.Conclusion: The machine-learning model, nomogram and online calculator might be useful to access the onset of severe and critical illness among COVID-19 patients and triage at hospital admission.

The aim of this study was to describe clinical, imaging, and laboratory features of acute pulmonary embolism (APE) in patients with COVID-19 associated pneumonia. Patients with COVID-19 associated pneumonia who underwent a computed tomography pulmonary artery (CTPA) scan for suspected APE were retrospectively studied. Laboratory data and CTPA images were collected. Imaging characteristics were analyzed descriptively. Laboratory data were analyzed and compared between patients with and without APE. A series of 25 COVID-19 patients who underwent CTPA between January 2020 and February 2020 were enrolled. The median D-dimer level founded in these 25 patients was 6.06 μg/mL (interquartile range [IQR] 1.90-14.31 μg/mL). Ten (40%) patients with APE had a significantly higher level of D-dimer (median, 11.07 μg/mL; IQR, 7.12-21.66 vs median, 2.44 μg/mL; IQR, 1.68-8.34, respectively, P = .003), compared with the 15 (60%) patients without APE. No significant differences in other laboratory data were found between patients with and without APE. Among the 10 patients with APE, 6 (60%) had a bilateral pulmonary embolism, while 4 had a unilateral embolism. The thrombus-prone sites were the right lower lobe (70%), the left upper lobe (60%), both upper lobe (40%) and the right middle lobe (20%). The thrombus was partially or completely absorbed after anticoagulant therapy in 3 patients who underwent a follow-up CTPA. Patients with COVID-19 associated pneumonia have a risk of developing APE during the disease. When the D-dimer level abnormally increases in patients with COVID-19 pneumonia, CTPA should be performed to detect and assess the severity of APE.