### combined cut off value of 2 variables when combining their roc using predicted probability

Posted:

**Mon Mar 20, 2017 10:43 am**I wanted to combine two roc curves of two variables which gave me a better area under curve. This i did using predicted probability derived from binary logistic regression of the two variables. However from this new roc curve and coordinates i am not able to get the cut off value of the combined variables. I am only getting a predicted probability value at say 80% sesnsitivity and 80% specificty from the coordinates but the corresponding variable values is missing. How to arrive at that? Kindly help