Some quick thoughts:
(1) Surely some researcher has come up with a quantile regression model that gives better results than period/cohort table conditioned only on sex.
There are great calculators out there that take into account loads of demographic and medical information. But they're not free.