Washington: A new artificial intelligence (AI) approach has been developed that can identify whether a five-day-old in vitro fertilized human embryo can progress to a successful pregnancy. The technique analyzes time-lapse images of the early-stage embryos and could potentially enhance the success rate of in vitro fertilization (IVF) and minimize the risk of multiple pregnancies.

The average success rate of IVF in the US is about 45 per cent. In the study, published in NPJ Digital Medicine, 12,000 photos of human embryos, taken 110 hours after fertilization, were employed to train an AI algorithm to differentiate between poor and good embryo quality.

Each embryo was first assigned a grade by embryologists. A statistical analysis then correlated the embryo grade with the probability of proceeding to a successful pregnancy outcome.

Embryos were considered good quality if the chances were more than 58 per cent and poor quality if the chances were less than 35 per cent. Following training and validation, the algorithm classified the quality of a new set of images with 97 per cent accuracy.

Zev Rosenwaks, from Weill Cornell Medical College, mentioned that with the introduction of the new technology into the field of IVF, it will be possible to automate and standardize a process that was dependent on subjective human judgement.

Choosing the embryo that has the best chances of developing into a healthy pregnancy is currently a subjective process. Agreement is low on how to predict the viability of an embryo based on its appearance at the blastocyst stage.

Nikica Zaninovic, from the Weill Cornell Medicine, said that they wanted to develop an objective method for standardization and optimization of the selection process in order to enhance the success rates of IVF.

More than six months were spent reviewing around 50,000 anonymized images, depicting 10,148 human embryos, accumulated through time-lapse photography over seven years. Two sets of 6,000 images, good or poor quality, were used to train the algorithm how to classify new images presented to it.

Pegah Khosravi, the lead author of the study, said that this is the first time that a deep learning algorithm has been applied on human embryos with so many images.