Skip to main content

Table 3 AUROC and calibration measures for scores and formulas from selected articles with cut-offs

From: Clinical prediction models for diagnosis of COVID-19 among adult patients: a validation and agreement study

  qRT-PCR test
All Positive Negative Discrimination and calibration results (Rubin’s rules)
         Calibration
N P50 (P25–P75) N P50 (P25–P75) N P50 (P25–P75) AUROC (95CI%) Brier score Slope (95 CI%) Intercept (95 CI%)
Vieceli et al 1475   492   983      
 Score (0–9)   6 (4–7)   7 (6–9)   4 (4–6) 0.77 (0.74–0.79)    
Model        0.77 (0.74–0.89) 0.18 0.55 (0.48–0.62) − 0.00 (− 0.14 to 0.14)
Original value       0.85 (0.77–0.92)    
Tordjman et al 1618   519   1099      
 Score (0–5)   3 (2–4)   4 (3–5)   2 (1–4) 0.73 (0.71–0.76)    
 Model        0.74 (0.72–0.77) 0.19 0.40 (0.35–0.46) − 0.0 (− 0.14 to 0.14)
 Original value       0.92 (0.89–0.95)    
Kurstjens et al 1472   489   983      
 Score (0–14)   6 (3–14)   10 (7–12)   4 (2–6) 0.84 (0.83–0.87)    
 Original value       0.94 (0.91–0.96)    
Aldobyany et al 502   224   278      
 Score (0–8)   4 (3–5)   5 (3.5–6)   3 (2–5) 0.60 (0.57–0.63)    
 Original value       0.60 (0.57–0.64)    
Nakakubo et al 1615   519   1096      
 Score (0–11)   4 (3–6)   6 (5–6)   4 (3–5) 0.78 (0.76–0.81)    
 Original value       /    
Fink et al 564   259   305   0.78 (0.75–0.80) 0.18 1.13 (0.99–1.27) 0.00 (− 0.11 to 0.11)
 Model           
 Original value       0.85 (0.81–0.90)