TimelyScholarFree Tools › SROC Curve Generator
Pool sensitivity and specificity across diagnostic accuracy studies with the bivariate model, and plot the summary ROC curve with the summary point and its 95% confidence region.
FreeNo account neededYour data stays in your browser
| Summary sensitivity | 89.1% (95% CI 80.8 to 94.1) |
|---|---|
| Summary specificity | 78.0% (95% CI 71.5 to 83.3) |
| Between-study variance (logit) | sensitivity 1.380; false positive rate 0.407 |
| Correlation | 0.854Between logit sensitivity and logit false positive rate |
At least one study had a zero cell, so 0.5 was added to every cell of every study (as in mada).
The summary point (red) has a 95% confidence region; the SROC curve (orange) is the Rutter-Gatsonis curve implied by the bivariate model, drawn over the range of observed false positive rates. Results match the mada R package (reitsma, sroc).
Columns TP, FP, FN and TN, one row per study.
Pooled sensitivity, specificity and their correlation.
SVG or PNG for your manuscript.
Diagnostic accuracy studies report pairs of sensitivity and specificity that are usually negatively correlated, because studies use different thresholds. The bivariate model pools the two together and accounts for that correlation.
The summary point is the pooled sensitivity and specificity; its confidence region shows the uncertainty. The SROC curve shows how sensitivity and specificity trade off across the range of observed false positive rates.
Separate pooling ignores their correlation and threshold effects, which can give misleading summary estimates.