مقالات پذیرفته شده در نهمین کنگره بین المللی زیست پزشکی
The Application of Artificial Intelligence in the Prediction and Management of Infertility in Women with Polycystic Ovary Syndrome (PCOS): A Systematic Review
The Application of Artificial Intelligence in the Prediction and Management of Infertility in Women with Polycystic Ovary Syndrome (PCOS): A Systematic Review
Zahra Tajalifar,1,*Shohreh Najafi Pour,2Sima Nazarpour,3Fatemeh Tajalifar,4
1. B.Sc. Student in Operating Room Technology, Faculty of Paramedical Sciences, Varamin-Pishva Branch, Islamic Azad University, Varamin, Iran 2. M.Sc. in Marine Biology, Farabi Research Center, District 12 Ministry of Education, Tehran, Iran 3. Associate Professor, Department of Midwifery, VaP.C., Islamic Azad University, Varamin, Iran 4. B.Sc. in Microbiology, East Tehran Branch, Islamic Azad University, Tehran, Iran
Introduction: Polycystic Ovary Syndrome (PCOS), the most common endocrine disorder among women of reproductive age, is a leading cause of infertility. In the current situation, where the country faces serious challenges such as a declining population growth rate (1.2%) and an increasing share of the elderly (10.3%), addressing infertility barriers like PCOS has become increasingly crucial. This syndrome is characterized by hyperandrogenism, ovulatory dysfunction, and polycystic ovarian morphology. Despite recent advances in Assisted Reproductive Technologies (ART), accurately predicting treatment outcomes in women with PCOS remains a major challenge. In this context, Artificial Intelligence (AI), with its capacity to analyze complex and multidimensional data (including ultrasound images, hormonal biomarkers, and metabolic parameters), has emerged as a transformative tool in reproductive medicine. The first AI-based model for detecting endometrial patterns in PCOS was introduced by Lee et al. in 2019, with an accuracy of 82%. This systematic review was conducted to optimize infertility treatment in women with PCOS through AI-based approaches.
Methods: To conduct this systematic review, a comprehensive search was performed in PubMed, Scopus, Web of Science, and Embase using combinations of relevant keywords, including artificial intelligence, polycystic ovary syndrome (PCOS), infertility, machine learning, and convolutional neural networks. The inclusion criteria encompassed all types of studies, including descriptive, analytical, and clinical trials, published in both English and Persian. Exclusion criteria included conference abstracts, lectures, and studies lacking sufficient evidence. Only full-text articles relevant to the study objectives were selected. The search was limited to publications from 2017 to 2025. Additionally, the reference lists of selected studies were manually reviewed to identify further relevant articles. In total, 15 eligible studies were included in the final analysis.
Results: AI has created a significant transformation in predicting and managing infertility in women with PCOS. Convolutional Neural Network (CNN) algorithms demonstrated a sensitivity of 94% and a specificity of 89% in analyzing ovarian ultrasound images for the detection of cysts and ovarian volume. Data-driven intelligent systems based on biochemical markers successfully classified PCOS subgroups with an accuracy of 87%. These intelligent systems, by integrating ultrasound images and blood test results, provided more accurate and rapid diagnoses. Furthermore, the XGBoost model predicted the optimal clomiphene dose with 85% accuracy, leading to a 35% reduction in fertility drug side effects, a 22% increase in pregnancy rates, and more precise dose adjustments. Combining laboratory data with patients’ psychological status also improved treatment success prediction accuracy up to 84%, supporting physicians in choosing the most appropriate therapy.
Conclusion: The findings of this systematic review clearly indicate that Artificial Intelligence (AI), particularly Machine Learning (ML), is on the verge of creating a fundamental shift in the management of Polycystic Ovary Syndrome (PCOS) and its related infertility. The review confirmed the effectiveness of various AI models in different areas, such as accurate and rapid diagnosis and treatment outcome prediction. Integrating these technologies into healthcare systems can lead to earlier and more timely diagnosis of PCOS, reduced treatment-related complications, and, ultimately, improved pregnancy success rates in this patient population. This advancement represents not only a medical breakthrough but also an important step toward addressing demographic challenges.