INTRODUCTION: Polycystic Ovarian Disease (PCOD) is a complicated endocrine-metabolic disorder affecting about one-quarter of women of reproductive age in the world and a major cause of infertility. This disorder is characterised by hyperandrogenism, anovulation, insulin resistance and metabolic abnormalities that pose challenges for timely diagnosis and management. Standardised criteria and symptom variability often limit traditional diagnostic strategies. OBJECTIVE: This study aims to evaluate the role of artificial intelligence (AI) technologies in enhancing the diagnosis, prediction and management of PCOD. METHODS: The systematic literature review was performed following PRISMA guidelines and included studies from 2021 to 2025. We reviewed more than 140 peer-reviewed publications in the clinical, biochemical, imaging, and multi-omics domains. The review covers machine learning (ML), deep learning (DL), hybrid AI models, explainable AI (XAI), federated learning (FL), quantum machine learning (QML), Edge AI, and generative adversarial networks (GANs). RESULTS: The results demonstrate the superior performance of ML, DL, and hybrid AI frameworks compared to conventional diagnostic methods in PCOD classification and prediction of metabolic and reproductive risks. XAI provided transparency into the model, and FL facilitated privacy-preserving sharing of data from multiple institutions. QML and integration of multi-omics showed promise for biomarker discovery. The challenges of limited datasets and real-time screening were addressed through GAN-based augmentation and Edge AI. DISCUSSION: These findings underscore the growing clinical relevance of AI in enhancing diagnostic accuracy and facilitating personalised decision-making. However, routine clinical implementation is still hindered by limitations such as data heterogeneity and imbalance, limited external validation, and lack of standardised datasets. CONCLUSION: AI-based methods present enormous potential to revolutionise the diagnosis and management of PCOD by providing accurate, interpretable, and personalised care. Future work should be based on large multicentre datasets, standardised validation protocols, and development of clinically interpretable models.