AI System BELA Enhances Embryo Assessment for IVF Success
核心洞察
Researchers at Weill Cornell Medicine developed BELA, an AI-based platform, to assess embryo chromosomal status, a key factor in IVF success.
BELA uses time-lapse video images to predict embryo ploidy, eliminating the need for subjective embryologist assessments.
Testing showed BELA predicted ploidy status with higher accuracy than previous AI models across multiple IVF clinic datasets.
An AI-based platform named BELA (Blastocyst Evaluation via deep Learning Assessment) has been developed by researchers at Weill Cornell Medicine to objectively assess the chromosomal status of in vitro fertilization (IVF) embryos, a critical determinant of IVF success. Unlike previous AI approaches, BELA does not rely on subjective embryologist assessments, offering a more generalizable measure for predicting whether an embryo has a normal (euploid) or abnormal (aneuploid) number of chromosomes.
Objective Embryo Quality Assessment
Embryologists traditionally evaluate embryo quality through microscopic examination. However, in cases with suspected issues, such as advanced maternal age, preimplantation genetic testing for aneuploidy (搜索) (PGT-A), a biopsy-like procedure, is used as the 'gold standard' despite its risks. BELA aims to automate this workflow and improve outcomes by generating accurate ploidy predictions independent of embryologist input.
The core of BELA is a machine-learning model that analyzes nine time-lapse video images of an embryo during a key interval around five days post-fertilization. This analysis generates an embryo quality score, which, combined with maternal age, predicts euploidy or aneuploidy (搜索). In a 2022 study, a predecessor system called STORK-A achieved approximately 70% accuracy using a single microscopic image, maternal age, and embryologists’ scoring.
Model Training and Validation
The model was trained using a deidentified dataset of nearly 2,000 embryos and their PGT-A-tested ploidy status from Weill Cornell Medicine CRM. The model was then tested on both internal and external datasets from IVF clinics in Florida and Spain. Results indicated that BELA predicted ploidy status with 'moderately higher accuracy' than previous versions and performed well across different datasets.
Clinical Implications and Future Directions
Nikica Zaninovic, associate professor of embryology in clinical obstetrics and gynaecology at Weill Cornell Medicine, stated that "BELA and AI models like it could expand the availability of IVF to areas that don’t have access to high-end IVF technology and PGT testing, improving equity in IVF care across the world."
The next step involves a prospective, randomized, controlled clinical trial to validate BELA’s predictive power. This research, supported in part by the National Institute of General Medical Sciences, was published in Nature Communications.