Model parameters
| Study | Model construction | Features | Predicted outcome | Data processing |
| Gong et al5 | Nomogram built from LASSO algorithm and logistic regression model | Age, direct bilirubin, BUN, CRP, LDH, albumin, RDW | Shortness of breath, respiratory rate ≥30/min, SpO2 ≤93% or PaO2/FiO2 ≤300 mm Hg | Variables with >7% values missing excluded. Other missing variables imputed by expectation maximisation |
| Zhou et al8 | Formula derived from logistic regression | Neutrophil count, lymphocyte count, D-dimer | Respiratory rate ≥30/min, severe respiratory distress, SpO2 <90% on room air, ARDS, sepsis or septic shock | Not discussed |
| Zou et al9 | Nomogram derived from logistic regression and Cox regression model | Age, disturbance of consciousness (GCS under 15,), LDH, CRP, chronic heart disease, chronic renal insufficiency, septic shock | In-hospital mortality | Not discussed |
| Xie et al5 | Nomogram derived from logistic regression | Age, LDH, log (lymphocyte count), SpO2 | In-hospital mortality | Not discussed |
| Yan et al6 | Multitree XGBoost | LDH, lymphocyte percent, CRP | In-hospital mortality | Pregnant, lactating women, minors, cases with data <80% complete excluded. Missing data ‘−1’ padded |
| Levy et al10 | LASSO regression, multivariate logistic regression | Age, BUN, Emergency Severity Index, RDW, neutrophil count, serum bicarbonate, serum glucose | In-hospital mortality | Not discussed |
| Zhang et al11 | Logistic regression | Age, sex, neutrophil count, lymphocyte count, platelet count, CRP, D-dimer, creatinine, ALT | Task 1: ARDS, intubation, ECMO, ICU admission or in-hospital mortality | Not discussed |
| Task 2: ARDS, intubation, ECMO, ICU admission or in-hospital mortality | ||||
| Guo et al12 | Cox regression | Age, underlying chronic disease, neutrophil-to-lymphocyte ratio, CRP, D-dimer | Shortness of breath, respiratory rate ≥30/min, SpO2 ≤93%, PaO2/FiO2 ≤300 mm Hg, respiratory failure with mechanical ventilation, circulatory shock, multiple organ or failure with ICU admission | Cases with missing data excluded |
| Hu et al13 | Logistic regression | Age, CRP, lymphocyte count, D-dimer | In-hospital mortality | Missing values imputed using bagging trees |
| Carr et al14 | Logistic regression augmented by XGBoost | NEWS2 (respiratory rate, SpO2, systolic BP, heart rate, GCS<15, temperature, supplemental oxygen (binary)), CRP, neutrophil count, eGFR, albumin, age | In-hospital mortality or ICU admission | Not discussed |
ALT, alanine aminotransferase; ARDS, acute respiratory distress syndrome; BP, blood pressure; BUN, blood urea nitrogen; CRP, C-reactive protein; ECMO, extracorporeal membrane oxygenation; eGFR, estimated glomerular filtration rate; FiO2, fraction of inspired oxygen; GCS, Glascow coma score; ICU, intensive care unit; LASSO, least absolute shrinkage and selection operator; LDH, lactate dehydrogenase; NEWS2, national early warning score 2; PaO2, partial pressure of oxygen; RDW, erythrocyte distribution width; SpO2, oxygen saturation.