Life Whisperer

Life Whisperer is developing a cloud and AI-based embryo-identification deep learning system to aid embryologists in the identification of the most viable embryos, thus increasing IVF certainty and allowing more patients to attempt in vitro fertilization (IVF). Life Whisperer is the fertility arm of Presagen, an AI healthcare company.

Life Whisperer tools were developed in collaboration with IVF clinics. The Life Whisperer software analyzes high-resolution digital camera images taken of Day 5 embryos to assist in identifying morphological features that correlate with uterine implantation potential. A confidence score, provided by the AI algorithm, objectively indicates an embryo’s implantation and viability potential. The AI model is incorporated into a cloud-based software application accessible via the web. Embryologists are able to upload images of embryos through computer or mobile device and the AI model instantly returns a viability confidence score.

Introduction to Life Whisperer.

Life Whisperer Viability assesses implantation potential and likelihood of clinical pregnancy. Life Whisperer Genetics assesses the likelihood of the embryo being chromosomally normal (euploid), as an alternative to invasive PGT-A genetic testing. These two products are CE-Marked and authorized to sell to two thirds of the world’s IVF market as of June 2021. The third IVF product is focused on oocyte evaluation.

To develop their image analysis model for prediction of human embryo viability, Life Whisperer used ensemble modeling to combine computer vision methods and deep learning neural network techniques. The Life Whisperer AI model was trained on images of Day 5 blastocysts at all stages including early, expanded, hatching, and hatched blastocysts. In the initial pilot study, an AI-based model surpassed the original objective and demonstrated an accuracy improvement of 30.8% over trained embryologists. Model accuracy was marginally lower on further development due to the introduction of inter-clinic variability. The final AI model showed an improvement of 24.7 percent over the predictive accuracy of embryologists for binary viable/non-viable classification.