AiVF, a Israeli startup that pairs AI with IVF, is introducing a new AI technology that reveals chromosomal abnormalities in human embryos growing in the IVF lab, without touching the embryo. This breakthrough can help women avoid miscarriage and increase their chances of delivering a healthy baby.
AiVF analyzed an enormous dataset—more than 2,500 embryos with known chromosomal makeup—using computer vision and deep neural networks, discovering a completely new perspective on biological development. In this study, time-lapse videos of 1,500 chromosomally normal and 1,000 chromosomally abnormal embryos were evaluated. The non-invasive AI-based algorithm successfully identified abnormal embryos in 73% of cases.
“Non-invasive, AI-based testing will completely replace the commonly used biopsy-based pre-implantation genetic testing, which is currently the only mode of testing,” said clinical embryologist and biostatistician Daniella Gilboa, CEO and co-founder of AiVF. “It ensures normal embryological development while reducing the emotional and financial strain on the parent.”
“AI technology is being used to distinguish between euploid and aneuploid embryos by computer-vision in an objective and indirect way based on the embryo metabolic activity,” said Dr. Marcos Meseguer of IVIRMA Spain, which is part of the research study. “Our joint research reveals that euploid and aneuploid embryos are visually distinct. This non-invasive PGT system could conceivably tell us more than current invasive methods without the cost and damage to the embryo that these incur.”
Existing procedures require a significant investment of both time and money. The expensive invasive testing process of pre-implantation genetic testing (PGT) requires the mother to wait an additional month for embryo transfer.
“With an ever-increasing volume of genetic and genomic information and clinical correlations that better delineate molecular mechanisms of disease and development, the need for AI to characterize this information has never been greater,” explained Dr. Lee P. Shulman, Professor of Obstetrics and Gynecology, Feinberg School of Medicine, Northwestern University.
The AiVF testing takes place in real-time, with results observed second-by-second as embryos develop. If the embryos are found to be healthy, they can be transferred immediately.
AiVF’s innovative tool will increase the chances of achieving a normal pregnancy. AiVF offers a global assessment of the entire embryo while other methods are based on a limited sample of a few cells that might not be representative of the complete embryo, a problem known as mosaicism.
AiVF has the largest database in the world of embryonic development, outcomes, genetic testing, and genomic data. This is the first time a non-invasive technology based on digital evaluation of time-lapse videos demonstrated that embryos with an abnormal genotype can be identified based on their phenotype, i.e., appearance over time.
“The value of AI in helping to interpret cellular developmental mechanisms is best seen with its adaption for the prediction of embryonic outcomes in IVF processes,” explained Dr. Shulman. “AI assessment can provide an accurate and noninvasive evaluation of embryo quality and potential that will change our approach to IVF and improve outcomes.”
AiVF is currently developing a more advanced AI system using a multilayer 3D architecture that is expected to significantly increase sensitivity for detecting chromosomally abnormal human embryos. The study is supported by a European Union Eureka grant and was done in collaboration with the leading IVF center Instituto Valenciano De Infertilidad (IVIRMA).