Diagnostic Accuracy Calculator
Sensitivity, specificity, PPV, NPV, LR± and 95% confidence intervals.
Explore the 1,000-person prevalence simulator →Ready when you are.
Enter your values and select Calculate to see your result here.
Analyze a 2×2 diagnostic classification table, including Wilson 95% confidence intervals, likelihood ratios, diagnostic odds ratio, F1 and prevalence-adjusted predictive values.
How it works
Method: Sensitivity = TP/(TP + FN), specificity = TN/(TN + FP); LR+ = sensitivity/(1 − specificity), LR− = (1 − sensitivity)/specificity. Youden J = sensitivity + specificity − 1. F1 = 2TP/(2TP + FP + FN). Wilson score intervals are supplied only for proportions.
Example: With TP = 85, FN = 15, FP = 10 and TN = 90, sensitivity is 85%, specificity 90%, LR+ 8.5 and LR− ≈ 0.167. Population PPV/NPV can be estimated for a given prevalence using Bayes’ theorem.
Statistical references
Wilson EB. Probable inference, the law of succession, and statistical inference. J Am Stat Assoc. 1927. DOI. See also STARD 2015 for reporting diagnostic accuracy studies.
1,000 people. How many test positive?
Move the prevalence slider to see how the same test creates different numbers of true and false positives. First click Calculate above to use your test’s sensitivity and specificity.
PPV and NPV use continuous expected probabilities. Dots and TP/FP/FN/TN counts are rounded illustrations of the expected numbers per 1,000 and can differ slightly from those probabilities. This is not a clinical study or a prediction for an individual.
Cite this toolCopy · BibTeX · RIS
If you use this calculator in research, you may cite the software and report the statistical methods separately. Citation is appreciated, not required. No DOI is registered for this release.
Reproducibility: download the version 1.6 source and tests · SHA-256 checksum. Future updates must retain this archive.
A citation cannot replace a methods description or independent verification. Report version, statistical assumptions and data-processing choices.