Combining Initial Radiographs and Clinical Variables Improves Deep Learning Prognostication in Patients with COVID-19 from the Emergency Department

Abstract

“Just Accepted” papers have undergone full peer review and have been accepted for publication in Radiology: Artificial Intelligence . This article will undergo copyediting, layout, and proof review before it is published in its final version. Please note that during production of the final copyedited article, errors may be discovered which could affect the content. Purpose To train a deep learning classification algorithm to predict chest radiography severity scores and clinical outcomes in patients with coronavirus disease 2019 (COVID-19). Materials and Methods In this retrospective cohort study, we identified patients of ages of 21 to 50 who presented to the emergency department (ED) of a multicenter urban health system from March 10–26, 2020 with COVID-19 confirmation on real-time reverse transcription polymerase chain reaction. We collected the initial chest radiographs (CXRs), clinical variables, and outcomes including admission, intubation, and survival within 30 days ( n = 338; median age 39; 210 men). Two fellowship-trained cardiothoracic radiologists examined CXRs for opacities and assigned a clinically validated severity score. We trained a deep learning algorithm to predict outcomes on a holdout test set composed of confirmed COVID-19 patients who presented from March 27–29, 2020 ( n = 161; median age 60; 98 men) for both younger (ages 21–50; n = 51) and older (ages > 50; n = 110) populations. Bootstrapping methods computed confidence intervals. Results The model trained on the CXR severity score produced the following areas under the receiver operating characteristic (AUCs): 0.80 (0.73,0.88) for the CXR severity score, 0.76 (0.68,0.84) for admission, 0.66 (0.56,0.75) for intubation, and 0.59 (0.49,0.69) for death. The model trained on clinical variables produced the following AUCs 0.64 (0.55,0.73) for intubation and 0.59 (0.50,0.68) for death. Combining CXR and clinical variables increased AUC of intubation and death to 0.86 (0.79,0.96) and 0.82 (0.72,0.91), respectively. Conclusion The combination of imaging and clinical information improves outcome predictions.

Distinct cytokine profiles associated with COVID-19 severity and mortality

“…The odds ratio (89.6) and relative ratio (24.3) obtained here are higher than those previously reported for individual inflammatory cytokine levels with a maximum hazard ratio of 4.2, 26 genetic markers with maximum odds ratio of 2.5, 30 or chest radiography with maximum accuracy of 74%. 31 Our results emphasize the importance of cytokine combination profiling in assessing severity and mortality.…”

The negative impact of obesity on the occurrence and prognosis of the 2019 novel coronavirus (COVID-19) disease: a systematic review and meta-analysis

“…1 ); some studies reported data on > 1 pertinent outcome. Of these, there were 52 studies reporting results on obesity prevalence among patients with COVID-19 involving a total of 504,556 cases [ 8 , 9 , 16 – 25 , 31 – 70 ] and data regarding the secondary outcomes were reported in 43 studies [ 8 , 9 , 16 – 25 , 31 , 32 , 34 – 38 , 42 , 43 , 45 – 55 , 57 – 61 , 64 , 67 , 68 , 70 – 72 ]. A study by Chao et al [ 73 ] was exclusively performed on children and thus was not included in the meta-analysis.…”

Cross-sectional analysis of follow-up chest MRI and chest CT scans in patients previously affected by COVID-19

“…In the setting of COVID-19, it is a frequently performed examination since it is able to show signs of interstitial lung disease (ILD), peripheral consolidation, and acute respiratory distress syndrome [ 7 ], although it remains an operator-dependent assessment and it is not considered a sensitive test for the diagnosis of COVID-19. CT has already been demonstrated to be a highly sensitive—albeit not specific—modality for the diagnosis of COVID-19 [ 7 , 8 ] and CT findings for the disease have been reported to follow a relatively typical temporal pattern and make it possible to monitor the evolution of lung involvement during the clinical course of the condition [ 9 – 12 ].…”