EUVIMEDEUVIMED entdecken
KohortenstudieOpen Access · CC BY-NC-ND 4.0

Outcome-specific prognostic factors after vNOTES sacrocolpopexy: A single-center retrospective cohort study using machine learning with repeated cross-validation

Shuozhen Chen, Xingnan Zhou, Lihong Han, Yuwei Li, Wenyu Liu, Xunwei Wu, Lian Tang, Yuejiao Zhang, Youjun Zhou, Hui Kuang, Nannan Sun, Juan Liu

Intelligent Medicine · 2026-11

DOI: 10.1016/j.imed.2026.11.005 ↗

Kurz gefasst

Kohortenstudie in der Zeitschrift Intelligent Medicine, 2026-11. Verfasst von Shuozhen Chen und 11 weiteren Autor:innen. Die Arbeit nennt 39 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: Guangdong Basic and Applied Basic Research Foundation, National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Vaginal natural orifice transluminal endoscopic surgery (vNOTES) sacrocolpopexy is increasingly used for pelvic organ prolapse, but factors associated with specific long-term outcomes remain insufficiently characterized. This study aimed to identify factors associated with postoperative outcomes after vNOTES sacrocolpopexy using a conservative machine-learning framework.

Methods

This single-center retrospective study included 70 patients who underwent vNOTES sacrocolpopexy at the Third Affiliated Hospital of Guangzhou Medical University from June 1, 2017, to August 31, 2021, with structured long-term follow-up. Patients were grouped separately according to seven binary postoperative outcomes: lower urinary tract symptoms (LUTS), anterior pelvic prolapse recurrence, middle/apical pelvic prolapse recurrence, posterior pelvic prolapse recurrence, mesh exposure, resurgery, and objective satisfaction status. Candidate predictors included demographic, obstetric, surgical, pelvic organ prolapse quantification (POP-Q), and symptom-related variables. A two-stage machine-learning workflow combining Boruta feature screening and elastic-net logistic regression was applied with repeated stratified cross-validation and held-out internal test evaluation. Class imbalance was addressed by upsampling within folds, and discrimination was evaluated using repeated stratified cross-validation (10 folds × 5 repeats) and a held-out test split. A factor was labeled high-confidence only if it was (i) Boruta-confirmed, (ii) retained with a nonzero coefficient in the tuned elastic-net model, and (iii) associated with an outcome model with a held-out test area under the receiver operating characteristic curve (AUC) > 0.5.

Results

The mean follow-up duration was 66.04 ± 12.81 months. Postoperative LUTS occurred in 31/70 patients, middle/apical pelvic prolapse recurrence in 3/70, and resurgery in 5/70. No observations were removed by complete-case filtering. Under the predefined criteria, associated factors were identified for three outcomes only. Body mass index (BMI) was identified for postoperative LUTS, operative time for postoperative middle/apical pelvic prolapse recurrence, and postoperative total vaginal length (TVL), menopausal status, and age for resurgery. The held-out test AUCs were 0.518 for postoperative LUTS, 0.571 for middle/apical pelvic prolapse recurrence, and 0.923 for resurgery. Direction-calibration analyses supported a risk-increasing interpretation for BMI and operative time, whereas the resurgery-associated factors required cautious interpretation because of rare events and uncertainty in direction.

Conclusion

BMI, operative time, postoperative TVL, menopausal status, and age were identified as outcome-specific associated factors under the predefined machine-learning criteria. These findings are exploratory and require validation in larger multicenter cohorts.

Förderung

  • Guangdong Basic and Applied Basic Research Foundation
  • National Natural Science Foundation of China

Zitieren

APA

Chen, S., Zhou, X., Han, L., Li, Y., Liu, W., Wu, X., Tang, L., Zhang, Y., Zhou, Y., Kuang, H., Sun, N., & Liu, J. (2026). Outcome-specific prognostic factors after vNOTES sacrocolpopexy: A single-center retrospective cohort study using machine learning with repeated cross-validation. Intelligent Medicine. https://doi.org/10.1016/j.imed.2026.11.005

Vancouver

Chen S, Zhou X, Han L, Li Y, Liu W, Wu X, et al. Outcome-specific prognostic factors after vNOTES sacrocolpopexy: A single-center retrospective cohort study using machine learning with repeated cross-validation. Intelligent Medicine. 2026. doi:10.1016/j.imed.2026.11.005

BibTeX
@article{Chen2026,
  author = {Chen, Shuozhen and Zhou, Xingnan and Han, Lihong and Li, Yuwei and Liu, Wenyu and Wu, Xunwei and Tang, Lian and Zhang, Yuejiao and Zhou, Youjun and Kuang, Hui and Sun, Nannan and Liu, Juan},
  title = {Outcome-specific prognostic factors after vNOTES sacrocolpopexy: A single-center retrospective cohort study using machine learning with repeated cross-validation},
  journal = {Intelligent Medicine},
  year = {2026},
  doi = {10.1016/j.imed.2026.11.005}
}
Kostenlos starten →