According to a study published in the latest issue of The Lancet · Digital Health, researchers from Cornell University School of Medicine in the United States have developed a new artificial intelligence (AI) algorithm, which can avoid the shortcomings of biopsy and non-invasive determine whether the number of chromosomes in IVF embryos is normal, with an accuracy rate of about 70%. Abnormal chromosome number, called aneuploidy, is the main reason why in vitro fertilization (IVF) embryos cannot be implanted or healthy pregnancy cannot be achieved. At present, one of the methods for detecting aneuploidy involves sampling embryonic cells similar to biopsy and gene detection. This method increases the cost of IVF process and is invasive to embryos. At present, doctors mainly use microscopes to evaluate whether embryos have significant abnormalities related to poor viability. In order to obtain information about chromosomes, doctors can also use a biopsy method called pre implantation aneuploidy gene detection (PGT-A). In the new research, the research team developed STORK-A algorithm as a potential alternative to PGT-A, or as a more selective way to decide which embryos should be tested for PGT-A. The new algorithm STORK-A will automatically "learn" to associate some features of the data with the possibility of aneuploidy by using the embryo microscope image, embryo quality score, mother's age and other information taken 5 days after fertilization. The research team trained STORK-A on a data set of 10378 blastocysts, whose ploidy status is known. They tested the algorithm on independent datasets and found quite accurate results, which proved the universality of STORK-A. According to the evaluation of researchers, the accuracy of this algorithm in predicting aneuploid and normal chromosome "euploid" embryos is close to 70%. The accuracy of STORK-A in predicting aneuploidy (complex aneuploidy) involving multiple chromosomes compared with euploid was 77.6%. They hope to eventually be able to predict aneuploidy in a completely non-invasive way using AI and computer vision technology. The new algorithm represents the progress made in reducing the risk of IVF embryo selection, reducing subjectivity, reducing costs and improving accuracy. Researchers say this is a good example of AI's potential to change medicine. In the past few years, artificial intelligence has quietly set off a new revolution in the field of life science. One of the most well-known applications is the "alpha folding" developed by Google, which can accurately predict the static three-dimensional structure of proteins using artificial intelligence. Not long ago, Chinese scientists went further and developed an artificial intelligence model that can predict the dynamic structure of proteins. The above research is a unique way to apply artificial intelligence to predict whether the number of chromosomes in IVF embryos is normal. No matter in which kind of research and application, AI has perfectly demonstrated its "excellence": improving efficiency, reducing costs, and being convenient and fast. (Liu Xinshe)