Constructing a predictive model for anti-Müllerian hormone levels in reproductive-age women based on transvaginal three-dimensional ultrasound.

Abstract

Purpose: This study aimed to evaluate the predictive value of transvaginal three-dimensional (3D) ultrasound imaging parameters for anti-Müllerian hormone (AMH) levels in women of reproductive age.

Methods: A retrospective study was conducted on 492 patients who underwent ovarian reserve function assessments at Hangzhou Traditional Chinese Medicine Hospital. The patients were randomly divided into a training set (n = 420) and a validation set (n = 72). All participants underwent transvaginal 3D ultrasound to measure antral follicle count (AFC), ovarian volume (OV), peak systolic flow velocity (PSV), and resistance index (RI). Morning fasting venous blood samples were collected for AMH level measurement. Pearson correlation analysis was used to assess the relationships between age, AFC, OV, PSV, RI, and AMH. Multiple linear regression analysis was used to construct a unified regression model and a grouped regression model. The performance of the models was evaluated by comparing the root mean squared error (RMSE) and mean absolute error (MAE) between predicted and actual values.

Results: AMH showed a strong negative correlation with age and a positive correlation with OV, AFC, and PSV. AMH was significantly correlated with AFC and OV (r = 0.844 and 0.759, respectively; both P < 0.05), but only weakly correlated with PSV (r = 0.176, P < 0.05). The RMSE and MAE of the grouped regression prediction model were lower than those of the unified regression model.

Conclusion: The grouped regression AMH prediction model developed through multiple linear regression analysis demonstrated favorable performance, enabling accurate individualized prediction of AMH levels, thereby assisting clinicians in evaluating the fertility potential of women of reproductive age.