Fig. 2

This figure shows that women ranked higher by the predictive model are significantly more likely to have ovarian cancer. The predictive model ranked all 4669 women in the dataset based on their likelihood of having ovarian cancer, based solely on germ line DNA data. This ranking was then split into 50 equal partitions, each with about 93 women. This plot shows the odds ratio (relative to 414 ovarian cases out of 4669 total) of each of the 50 equal partitions along with the 95% confidence intervals