# What are ROC curves?

A ROC curve is a plot showing graphically the ability of a continuous variable to classify a binary variable (Yes-no) in either "Yes" or "No". It represents for every threshold of the continuous variable the sensitivity and the specificity for classifying patients according to the binary variable.

For example, blood glucose levels could be used to diagnose diabetes. For each value of blood glucose level, we can test if is able to detect diabetes and to eliminate non-diabetes. In this case, the ROC curve will be representing the sensitivity and specificity (actually 1-specificity) for each possible level of blood glucose.

# How to generate ROC curves and AUC?

- Go to "Test variables"
- Pick a Yes-no (binary) variable (for example "Diabetes")
- Pick a Numeric variable (for example "blood glucose")
- Click on the "ROC curves" panel

Several ROC curves templates will be displayed as well as the area under the curve (AUC).

# How to compare several ROC curves?

To compare several ROC curves, you only need to add more than one Numeric variable to the set of testes variables.

- Go to "Test variables"
- Pick a Yes-no variable
- Pick a first Numeric variable
- Pick one or several other Numeric variable(s)
- Click on the "ROC curves" panel
- A statistical test is automatically used to compare ROC curves AUC: the DeLong's test

# How to calculate sensitivity and specificity for a given threshold?

- Generate one or several ROC curves as described above
- Click on the panel "Sensitivity and Specificity"

Sensitivity and specificity are provided for each threshold of the Numeric variable.

Was this article helpful?

That’s Great!

Thank you for your feedback

Sorry! We couldn't be helpful

Thank you for your feedback

Feedback sent

We appreciate your effort and will try to fix the article