Abstract: Infertility remains a global health concern, and in-vitro fertilization (IVF) has become a widely used assisted reproductive technology for achieving pregnancy. However, predicting IVF success continues to be a major challenge due to biological variability, subjective embryo evaluation, and limited data integration. To address these issues, recent research has adopted Artificial Intelligence and Machine Learning paradigms to enhance prediction accuracy and automate embryo assessment. This study synthesizes and extends five advanced AI-based approaches that integrate deep learning, transformer architectures, and multi-modal data fusion for embryo grading and live birth outcome prediction. The unified framework leverages clinical, morphological, and temporal embryo features, applying models such as Convolutional Neural Networks (CNNs), Particle Swarm Optimization (PSO), and Tab-Transformers to extract interpretable and clinically relevant patterns. Comparative analysis shows that AI-driven systems can achieve accuracy levels exceeding, outperforming traditional embryologist evaluations. By providing explainable, data-driven insights, these methods have the potential to improve decision-making, reduce human subjectivity, and personalize IVF treatment outcomes.

Keywords: In-Vitro Fertilization, Artificial Intelligence, Machine Learning, Deep Learning, Transformer Models, Embryo Grading, Outcome Prediction, Multi-Modal Data Fusion, Explainable AI, Clinical Decision Support.

Downloads: | DOI: 10.17148/IARJSET.2025.121011