News | Artificial Intelligence in IVF: Reducing Costs and Risks
Researchers at Weill Cornell Medicine have developed an artificial intelligence algorithm called STORK-A that can noninvasively predict whether an in vitro fertilization (IVF) embryo has a normal chromosome count with about 70% accuracy. The advance may improve IVF success rates, reduce costs, and limit invasive embryo testing.
An abnormal number of chromosomes, known as aneuploidy, is a major reason IVF embryos fail to implant or develop into healthy pregnancies. Current aneuploidy testing typically requires a biopsy-like embryo sample and genetic testing, which increases IVF costs and is invasive to the embryo. STORK-A analyzes microscopic embryo images together with information such as maternal age and the IVF clinic’s embryo morphology score to predict aneuploidy without a biopsy.
STORK-A was trained on 10,378 blastocysts with known chromosome status. It distinguished aneuploid embryos from embryos with a normal chromosome count, known as euploid embryos, with nearly 70% accuracy (69.3%). Its accuracy was 77.6% when distinguishing complex aneuploidies involving multiple chromosomes from euploid embryos. Tests on independent datasets, including data from an IVF clinic in Spain, produced similar accuracy and demonstrated the algorithm’s generalizability.
Although STORK-A remains experimental, its development is an important step toward safer, more objective, affordable, and accurate IVF embryo selection. Future research will use videos of embryo development in an effort to improve STORK-A’s predictive accuracy.
News | Artificial Intelligence in IVF: Reducing Costs and Risks
News | Artificial Intelligence in IVF: Reducing Costs and Risks
Researchers at Weill Cornell Medicine have developed an artificial intelligence algorithm called STORK-A that can noninvasively predict whether an in vitro fertilization (IVF) embryo has a normal chromosome count with about 70% accuracy. The advance may improve IVF success rates, reduce costs, and limit invasive embryo testing.
An abnormal number of chromosomes, known as aneuploidy, is a major reason IVF embryos fail to implant or develop into healthy pregnancies. Current aneuploidy testing typically requires a biopsy-like embryo sample and genetic testing, which increases IVF costs and is invasive to the embryo. STORK-A analyzes microscopic embryo images together with information such as maternal age and the IVF clinic’s embryo morphology score to predict aneuploidy without a biopsy.
STORK-A was trained on 10,378 blastocysts with known chromosome status. It distinguished aneuploid embryos from embryos with a normal chromosome count, known as euploid embryos, with nearly 70% accuracy (69.3%). Its accuracy was 77.6% when distinguishing complex aneuploidies involving multiple chromosomes from euploid embryos. Tests on independent datasets, including data from an IVF clinic in Spain, produced similar accuracy and demonstrated the algorithm’s generalizability.
Although STORK-A remains experimental, its development is an important step toward safer, more objective, affordable, and accurate IVF embryo selection. Future research will use videos of embryo development in an effort to improve STORK-A’s predictive accuracy.
Story source:
Collected online