Researchers at Weill Cornell Medicine have introduced BELA, an AI-based system that uses time-lapse video images and maternal age to assess IVF embryo quality accurately, offering a promising future for improving IVF success rates.
Researchers at Weill Cornell Medicine have unveiled a groundbreaking artificial intelligence-based system called BELA, designed to assess the chromosomal status of in vitro-fertilized (IVF) embryos with remarkable precision. This innovative system utilizes time-lapse video images of embryos and maternal age, revolutionizing the traditionally subjective and labor-intensive processes of embryo evaluation.
Published in the journal Nature Communications, the study describes BELA as the team’s latest AI-based platform aiming to distinguish between embryos with a normal (euploid) or abnormal (aneuploid) number of chromosomes — a crucial factor for IVF success.
Unlike previous AI-based methods, BELA operates independently of embryologists’ subjective assessments, offering a fully automated and objective approach to embryo evaluation.
“This is a fully automated and more objective approach compared to prior approaches, and the larger amount of image data it uses can generate greater predictive power,” senior author Iman Hajirasouliha, an associate professor of physiology and biophysics at Weill Cornell Medicine, said in a news release.
Embryologists currently scrutinize IVF embryos under a microscope to assess their quality and determine any potential chromosomal issues, especially in cases of advanced maternal age. The “gold standard” test for this, called preimplantation genetic testing for aneuploidy (PGT-A), is a biopsy-like procedure involving some risk. Innovation in AI has become a vital avenue to minimize these risks and improve efficiency.
Hajirasouliha and his team had previously developed an AI-based system named STORK-A in 2022. This system used a single microscopic image of an embryo, alongside maternal age and embryologists’ scoring, which led to around 70% accuracy in predicting the embryo’s chromosomal status.
BELA improves on this by eliminating the need for embryologists’ input, relying solely on machine-learning algorithms to predict embryo quality with more accuracy.
The research team trained BELA using deidentified data from Weill Cornell Medicine’s Center for Reproductive Medicine (CRM) of nearly 2,000 embryos tested with PGT-A. They validated BELA’s performance with both internal and external datasets, including data from clinics in Florida and Spain, achieving higher accuracy than previous models.
The researchers’ next step is “to test BELA’s predictive power prospectively in a randomized, controlled clinical trial, which we are currently planning,” according to the news release.
“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,” added Nikica Zaninovic, an associate professor of embryology in clinical obstetrics and gynecology at Weill Cornell Medicine who led the embryology work for the study.
BELA’s capability to process vast amounts of embryo image data hints at its potential for broader applications in embryology.
“Our hope is that this model could be useful also for general embryo quality estimation, prediction of the embryo development stage, and other functions that an embryology clinic could tailor for its own needs,” added first author Suraj Rajendran, a doctoral student in Hajirasouliha’s laboratory.
The study emphasizes how AI can transform reproductive medicine by making it more efficient, precise and accessible, ultimately enhancing the chances of success for hopeful parents undergoing IVF treatment. The ongoing advancements in AI and machine learning in health care underscore the potential for technology to drive improved outcomes and broaden access to quality medical care.