Ah, if I misread then from the results section of the linked paper:

For the malignancy prediction objective, the algorithm obtained an area under the receiver operating characteristic curve (AUC) of 0.91 (95% CI: 0.89, 0.93), with specificity of 77.3% (95% CI: 69.2%, 85.4%) at a sensitivity of 87%.

I haven't read the papers methods, but the data set size is small-ish for this sort of analysis.

The value of the false positive rate is that it lets you know the probability of a true-miss. Depending on the classification exercise, you may be concerned with false positives, where the consequence of a missed call is significantly greater than an unwarranted checkup from a human doctor.