Discriminant models for the prediction of postponed viral shedding time and disease progression in COVID-19

Abstract

Abstract Background COVID-19 infection can cause life-threatening respiratory disease. This study aimed to fully characterize the clinical features associated with postponed viral shedding time and disease progression, then develop and validate two prognostic discriminant models. Methods This study included 125 hospitalized patients with COVID-19, for whom 44 parameters were recorded, including age, gender, underlying comorbidities, epidemiological features, laboratory indexes, imaging characteristics and therapeutic regimen, et al. Fisher's exact test and Mann–Whitney test were used for feature selection. All models were developed with fourfold cross-validation, and the final performances of each model were compared by the Area Under Receiving Operating Curve (AUROC). After optimizing the parameters via L2 regularization, prognostic discriminant models were built to predict postponed viral shedding time and disease progression of COVID-19 infection. The test set was then used to detect the predictive values via assessing models’ sensitivity and specificity. Results Sixty-nine patients had a postponed viral shedding time (> 14 days), and 28 of 125 patients progressed into severe cases. Six and eleven demographic, clinical features and therapeutic regimen were significantly associated with postponed viral shedding time and disease progressing, respectively (p < 0.05). The optimal discriminant models are: y1 (postponed viral shedding time) = − 0.244 + 0.2829x1 (the interval from the onset of symptoms to antiviral treatment) + 0.2306x4 (age) + 0.234x28 (Urea) − 0.2847x34 (Dual-antiviral therapy) + 0.3084x38 (Treatment with antibiotics) + 0.3025x21 (Treatment with Methylprednisolone); y2 (disease progression) = − 0.348–0.099x2 (interval from Jan 1st,2020 to individualized onset of symptoms) + 0.0945x4 (age) + 0.1176x5 (imaging characteristics) + 0.0398x8 (short-term exposure to Wuhan) − 0.1646x19 (lymphocyte counts) + 0.0914x20 (Neutrophil counts) + 0.1254x21 (Neutrphil/lymphocyte ratio) + 0.1397x22 (C-Reactive Protein) + 0.0814x23 (Procalcitonin) + 0.1294x24 (Lactic dehydrogenase) + 0.1099x29 (Creatine kinase).The output ≥ 0 predicted postponed viral shedding time or disease progressing to severe/critical state. These two models yielded the maximum AUROC and faired best in terms of prognostic performance (sensitivity of78.6%, 75%, and specificity of 66.7%, 88.9% for prediction of postponed viral shedding time and disease severity, respectively). Conclusion The two discriminant models could effectively predict the postponed viral shedding time and disease severity and could be used as early-warning tools for COVID-19.

Keywords

• Prognostic discriminant model
• Postponed viral shedding time
• Disease progression
• COVID-19