pubmed.info
Drag tiles by their title, resize at the bottom edge.

ROC curve and AUC calculator – with DeLong test and optimal cutoff

Paste the status (1/0) and up to four measurements – the tile draws the ROC curves, calculates the AUC with DeLong confidence interval, finds the optimal cutoff and compares the tests. Free, no account; your data are not uploaded.

More about this tool

What you can do

  • Draw ROC curves for one to four markers measured in the same individuals, optionally with the chance diagonal, grid lines and the optimal cutoff marked.
  • Calculate the AUC with standard error, DeLong confidence interval and P value for AUC = 0.5.
  • Find the optimal cutoff by the Youden index, the shortest distance to the corner (0|1), a minimum sensitivity or a minimum specificity.
  • Get sensitivity and specificity with Wilson confidence intervals, PPV, NPV, likelihood ratios and the 2 × 2 counts at the cutoff – and the same figures for a cutoff of your own.
  • Compare the AUCs of several tests with the DeLong test for correlated curves.
  • Look through all cutoffs in a table.

How it works

  • One row per individual: status (1/0, also yes/no or case/control), then the measurement(s), separated by tab, semicolon or space – directly from Excel or SPSS. A header row with names is allowed; decimal comma and point are both accepted.
  • Up to 20,000 rows. Rows with missing values (empty or NA) are excluded.
  • The direction is detected automatically: if lower values indicate status 1, the tile reverses it and tells you. You can also set it yourself.

Export

SVG with real text and your input embedded, vector PDF (Latin script only) or PNG at 300 or 600 dpi in common journal column widths, in colour or black and white with line patterns. The results and all cutoffs are available as CSV, together with a suggested text for methods and results. Load the SVG or a saved project file to continue later.

Your data: nothing is transmitted to a server. Only this browser remembers your input and settings.

Frequently asked questions

What does the AUC mean?

It is the probability that a randomly chosen individual with status 1 has a higher value than a randomly chosen individual with status 0. 0.5 corresponds to chance, 1.0 to perfect separation.

Is the Youden cutoff the best cutoff?

It weights false positives and false negatives equally. For ruling out disease choose a minimum sensitivity, for confirming it a minimum specificity. A cutoff found in the same data usually overestimates sensitivity and specificity and should be validated on independent data.

How do I compare two diagnostic tests?

If both tests were measured in the same individuals, enter them as two columns: the DeLong test accounts for the dependence of the measurements. With several comparisons, correct for multiple testing.

pubmed.info sets no tracking cookies and builds no user profiles. Visits are counted anonymously (cookieless, shortened IP addresses). Ads from Google appear only according to your consent choice.