Biostatistics & EBM Toolkit

Diagnostic accuracy, treatment effect, sample size, and Bayesian updating

Enter 2×2 Contingency Table

True Positives (TP)

False Positives (FP)

False Negatives (FN)

True Negatives (TN)

Pre-test probability / prevalence (%)

Used for Bayesian PPV/NPV. Change this to see how prevalence affects performance.

Sensitivity

90.0%

TP / (TP+FN)

Specificity

90.0%

TN / (TN+FP)

Accuracy

90.0%

(TP+TN) / All

Prevalence

10.0%

Pre-test prob

LR+

9.00

Se / (1−Sp)

LR−

0.11

(1−Se) / Sp

PPV (Bayes)

50.0%

@ 10% prevalence

NPV (Bayes)

98.8%

@ 10% prevalence

PPV (sample)

81.8%

TP / (TP+FP)

NPV (sample)

94.7%

TN / (TN+FN)

SpIN / SnOUT: High Specificity → Positive result rules IN the diagnosis. High Sensitivity → Negative result rules OUT the diagnosis. LR+ >10 and LR− <0.1 are considered highly significant.