JMIR Publications

ABSTRACT

Background:

Previously, we reported a model for assessing ovarian reserve using four predictors: anti Müllerian hormone (AMH) level, antral follicle count (AFC), follicle stimulating hormone (FSH) level, female age (A), together as the AAFA model.

Objective:

To explore the possibility of establishing a model for predicting ovarian reserve using only three factors: AMH and FSH levels, and age status (the AFA model).

Methods:

Gonadotropin-releasing hormone (GnRH) antagonist-based ovarian simulation cycles in our reproductive center were collected retrospectively. Poor ovarian response with <5 oocytes retrieved was defined as an outcome variable. The AFA model was built using a multivariate logistic regression analysis on data from 2017, and data from 2018 were used to validate its performance. Measurements of the area under the curve (AUC), sensitivity, specificity, positive predictive value, and negative predicative value were used to evaluate the performance of the model. To rank the ovarian reserve of the whole population, we ranked the subgroups according to the predicted probability of poor ovarian response and further divided the 60 subgroups into four clusters, A-D, according to cut-off values consistent with AAFA model.

Results:

The AUCs of the AFA and AAFA models were similar for the same validation set, with values of 0.853 (95% confidence interval, CI, 0.841–0.865) and 0.850 (0.838–0.862), respectively. We further ranked the ovarian reserve according to their predicted probability of poor ovarian response calculated using our AFA model. The actual incidences of poor ovarian response and 95% CI in groups A–D in the AFA model were 0.037 (0.029–0.046), 0.128 (0.099–0.165), 0.294 (0.250–0.341), and 0.624 (0.577–0.669), respectively. The order of ovarian reserve from adequate to poor followed the order AD. The clinical pregnancy rate, live-birth rate, and specific differences in groups A to D were similar when predicted using the AFA and AAFA models.

Conclusions:

This AFA model for assessing the true ovarian reserve was easier, more cost-effective and more objective than our original AAFA model. Clinical Trial: NA.