Artificial intelligence in embryo selection: enhancing precision and overcoming traditional limitations in in vitro fertilization

Identification of embryos with the highest potential for successful implantation is a key step in in-vitro fertilization (IVF). Traditionally, embryologists visually grade embryos by assessing their morphology and developmental stages. However, these assessments can differ between embryologists (inter-observer variability) and even when the same embryologist reviews the same embryo again (intra-observer variability), leading to inconsistent grading and potential misjudgement of embryo grading. Recent advancements in artificial intelligence (AI) offer a more standardized and objective approach to human embryo grading. By using machine learning models, AI systems can analyze embryo images and detect subtle developmental patterns that may not be apparent through visual assessment alone. This review explores original research studies from 2012 to 2024, that developed AI-driven embryo assessment methods that apply machine learning models, such as Convolutional Neural Networks (CNNs), which are deep learning models, while excluding studies involving animal embryos and non-english papers. Our findings from the review indicate that AI can reduce human error and improve embryo grading consistency for successful IVF. However, integrating AI into clinical practice presents challenges such as data variability, regulatory barriers, and the need for transparent, explainable AI models. Future directions include refining AI models to handle diverse datasets ensuring model interpretability for clinicians, and validating AI systems through large-scale clinical trials to establish their reliability and clinical utility in embryo selection.