Artificial intelligence-assisted controlled ovarian stimulation in in vitro fertilization: A critical narrative review.

Authors

Abstract

Artificial intelligence (AI) is increasingly being investigated as a decision-support tool in controlled ovarian stimulation (COS) during in vitro fertilization. Machine learning and deep learning algorithms have demonstrated promising predictive performance for outcomes including ovarian response, mature oocyte yield, premature luteinizing hormone rise, gonadotropin dose adjustment, and optimization of trigger timing. Convolutional neural networks applied to three-dimensional ultrasound imaging have also enabled automated follicular monitoring and more standardized assessment of follicular development. This narrative review summarizes current AI applications in COS and critically evaluates the methodological quality, validation status, and clinical relevance of available models. Most published studies are retrospective and based on single-center data sets, with limited external validation and scarce prospective randomized evidence. Furthermore, many reported improvements relate primarily to surrogate laboratory outcomes rather than clinically meaningful endpoints such as live birth rates. Although AI-assisted COS demonstrates considerable potential for individualized treatment and workflow optimization, important barriers remain, including model interpretability, generalizability, ethical considerations, data privacy, and regulatory oversight. Future progress will require multicenter collaborations, prospective clinical validation, explainable AI frameworks, and careful integration of AI tools into clinician-guided reproductive care.

Tags

Topics