Reference sample vials beside two clearly separated result trays. Original AI-generated artwork produced for GreenCalc; not a measured diagram or empirical data display.
Define the actual label before estimating test performance
Sensitivity is the fraction of actual positives that test positive. Specificity is the fraction of actual negatives that test negative. “Actual” must refer to a defined reference standard or outcome-assessment process. If that standard misclassifies cases, the resulting table measures agreement with it rather than unquestionable truth.
Use the appropriate denominator for each rate
With TP = 80, FN = 20, TN = 90 and FP = 10, sensitivity is 80/(80+20) = 80%. Specificity is 90/(90+10) = 90%. Predicted-positive count TP+FP is not the denominator of either rate; it belongs to precision, a different question.
Inspect how validation cases were assembled
A dataset containing only easy positive cases and obvious negatives may not represent the intended application. Different thresholds, measurement conditions or case mixes can change observed rates. A large number of observations does not by itself repair a biased selection process or an unsuitable reference standard.
Report the supporting counts
The Sensitivity and Specificity Calculator reports sample rates, their denominators and likelihood ratios without confidence intervals. A group with no actual positives leaves sensitivity undefined, not zero. Read LR plus and LR minus for the evidence-ratio interpretation. The existing Confusion Matrix Calculator provides the broader classifier metric report; the Post-Test Probability Calculator instead updates an entered pre-test probability.
Check the model behind your result
Use the Sensitivity and Specificity Calculator for the supported calculation and review its stated assumptions before interpreting the output. Find related tools in the Statistics category, or browse the GreenCalc homepage for other calculation topics.
By GreenCalc Editorial Team · Updated 2026-10-26