Create AUC-ROC from single sensitivity and specificity value? Is it possible and appropriate to estimate the area under the receiver operating characteristic curve from a single point estimate of an individual's sensitivity and specificity performance?
 A: NOT REALLY
That gives you one point (call it $(x,y)$) in the unit square $[0,1]\times[0,1]$. Think about how many non-decreasing curves can go from $(0,0)$ to that $(x,y)$ point to $(1,1)$.
One curve could be a diagonal line from $(0,0)$ to your $(x,y)$ point and then another diagonal line from your $(x,y)$ point to $(1,1)$, while another could start the same way while curving up in a quarter-circle path from the $(x,y)$ point to $(1,1)$. These have different areas.
EDIT
“Estimate” is a strange term, because there is a sense in which fixed numbers can be acceptable estimators (technical term is “admissible”), even if they are independent of the data, and nothing even requires us to pick an admissible estimator. Thus, a lot of values could be estimators of AUC, and you are correct to point out in the comments that you get a lower bound of sensitivity times specificity. That’s probably a weak lower bound, weak enough to be worthless, but I suppose it’s a better lower bound than $1/2$ or zero.
(Consider the lower bound you get if sensitivity and specificity each are $0.7$.)
