Artificial is giving birth to new forms of innovation, quite literally. In a new study, AI has assisted in the prediction of which embryos will be grown into healthy babies. The AI system was developed by scientists and its accuracy has been proven to be much more reliable and consistent than human embryologists.
The normal procedure to check the health of the embryo and to predict whether it will result in a healthy baby was traditionally carried out by an embryologist. This involves them checking the eggs by eye to see which of them appear to be healthy. Scientists found that they could teach AI to decipher the difference between healthy embryos and unhealthy embryos by showing the AI pictures. The test was first carried out with baby cows and now, by removing the potential for human error, IVF cycles can now result in healthier babies.
According to a report by The Independent, the technology was developed by experts from Sao Paulo State University in Brazil in a collaboration with the Boston Place Clinic in London along with Dr Cristina Hickman, the scientific adviser to the British Fertility Society. The technique has proven to be an inexpensive and more effective means of getting women pregnant sooner. The AI is able to distinguish 24 image-based characteristics of the embryo that are otherwise invisible to the human eye. Some of these characteristics include the size of the embryo, the texture of the image and the other biological features present such as the homogeneity of the cells. With this kind of a breakthrough, expecting parents will not need to waste time and money in treatments that would not work and can directly be given a proper examining of the embryo’s chances of developing into a healthy baby.
About 48 images of embryos were evaluated by embryologists and the AI and while the embryologists could not come to an agreement between the images, the AI had no problem in coming up with a solid and accurate prediction. What’s more is, the more the computer sees and executes this treatment, the more it will take back and learn again. As of now, there is still more work required for the usage of the AI to be conducted to its best capacity, but the experts are confident that the system will be available sooner than expected for clinical practice.