28 June 2018
Artificial Intelligence pioneered by Australian scientists to increase success rates in IVF pregnancies
Australian scientists are pioneering an innovative technology using artificial intelligence (“AI”) to predict the likelihood of a viable pregnancy prior to transferring an embryo in a woman undergoing IVF.
Called Ivy this artificial intelligence is an exciting development which allows embryologists to identify the embryo with the best chance of achieving a successful pregnancy as quickly as possible.
Dr Simon Cooke, Scientific Director at IVFAustralia, part of Virtus Health, Australia’s leading provider of assisted reproductive services and Aengus Tran, Chief Data Scientist at Harrison-AI, led the development of this technology.
The accuracy of predicting the likelihood that an embryo will result in a viable pregnancy can be increased using AI. Time-lapse videos of thousands of embryos during their development were used to ‘train’ the Ivy AI without the potential for human bias or subjective assessment.
Associate Professor Peter Illingworth, IVFAustralia’s Medical Director said: “The introduction of Ivy AI technology has the potential to contribute to IVF medicine by shortening the time frame to a successful pregnancy and increasing IVF pregnancy rates as it provides an EmbryoScore predicting an embryo’s potential to develop a fetal heart. We believe it is the most advanced tool of its kind in human embryology. Artificial Intelligence technology is likely, in future, to be the dominant embryo selection method used in IVF.”
By performing a comprehensive three-dimensional assessment of the growth of embryos through all stages of development and then relating this data to the corresponding pregnancy outcomes, Ivy has taught itself to identify those embryos with the highest potential of developing a fetal heart and allocates them with an ‘EmbryoScore’. The embryo with the highest score can then be selected and transferred, accelerating the chance of a healthy baby.
Currently, embryologists manually assess each embryo based on the physical appearance at a limited number of critical development check points using standard grading systems in conjunction with digital time lapse imagery to select the best embryo to transfer.
“Now with the Ivy AI technology we can take advantage of the ability of AI to process an enormous amount of data, too much for a human to process, to more accurately and objectively predict the best embryo to select for transfer, based on the presence of a fetal heart,” said A/Prof Illingworth.
Importantly, Ivy AI assessment of embryo time-lapse videos is non-invasive and risk-free as it does not require cells to be taken from the developing embryo. The AI observes the embryo’s development while the embryo remains completely undisturbed in the IVF time-lapse latest generation incubator.
Ivy can review and assess data from time-lapse videos of around five embryos per second.
Pre-clinical validation of the technology has been conducted in Virtus Health clinics using the data obtained from over 2600 embryos from Virtus Health’s NSW, ACT and Queensland laboratories. With a patent application lodged, Virtus Health will further evaluate Ivy and its EmbryoScore in a multicentre randomised clinical trial across the company’s Australian and European laboratories later this year enabling its rapid introduction to patient care.
“Ivy is a self-improving AI that continuously learns from embryos that it analyses. The AI iteratively retrains on powerful supercomputers and has been shown to consistently improve on previous versions of itself in identifying the embryo with the highest potential for a successful pregnancy” said Aengus Tran, Chief Data Scientist at Harrison-AI and final year medical student at the University of New South Wales. “As with many other areas of medicine, this is a demonstration of the potential for artificial intelligence to make an enormous contribution to human health.”