Interpretable machine learning prediction of live birth after freeze-all FET cycles across transfer-order subgroups - Takeaways - MDSpire

Interpretable machine learning prediction of live birth after freeze-all FET cycles across transfer-order subgroups

  • By

  • Yu Zhao

  • He Wang

  • Lin Wang

  • Lei Yan

  • Jiao Liu

  • Mengyi Teng

  • Hao Wang

  • Ting Liu

  • July 17, 2026

Share

  • 1

    The study developed predictive models for live birth outcomes after freeze-all FET cycles, focusing on transfer-order subgroup analysis.

  • 2

    Logistic regression, support vector machine, random forest, XGBoost, LightGBM, and CatBoost models were compared across three cohorts.

  • 3

    CatBoost with T6 and T10 feature templates showed the highest AUC values for the overall and second-transfer subgroups, respectively.

  • 4

    Key predictors identified included female age, ovarian-reserve indicators, basal progesterone, and interval days for the second-transfer subgroup.

  • 5

    Transfer-order subgroup modeling enhanced prediction accuracy and clinical interpretability for pre-transfer live-birth assessments.

Original Source(s)

Interpretable machine learning prediction of live birth after freeze-all FET cycles across transfer-order subgroups

  • by Yu Zhao, He Wang, Lin Wang, Lei Yan, Jiao Liu, Mengyi Teng, Hao Wang, Ting Liu

  • July 17, 2026

Related Content

Medium-sized follicle proportion on the trigger day may be associated with higher live birth rate in fresh embryo transfer cycles among low-prognosis patients: a retrospective cohort study

Occult cervical microinvasive squamous cell carcinoma with superficial endometrial spread mimicking high-grade squamous intraepithelial lesion in a postmenopausal woman: a case report

Case Report: A diagnostic challenge: rare adverse effects of GnRH analogue therapy in a late pubertal pediatric patient