Computational Model Combines Genetic, Clinical Features to Improve IVF

Using both clinical parameters and genetic characteristics to help predict the number of egg cells collected during in vitro fertilization (IVF) can improve the safety and the efficacy of the procedure, according to a study in this week's PLOS Computational Biology. IFV is a well-established infertility treatment, but many factors — including genetic ones — can affect treatment success, making outcome prediction difficult. One aspect of the IVF process that is particularly challenging to determine is the number of viable, fertilization-ready egg cells that can be collected after ovarian stimulation. To help predict this, a team led by scientists at Poland's Invicta Fertility Clinic built a machine learning model to identify features that could predict how many egg cells will be obtained during IVF. They find that anti-Müllerian hormone levels and antral follicle count were the most important clinical parameters to egg cell count prediction, with sequence variants in five reproduction-related genes as the most important genetic predictors. The model, the study's authors write, should help improve IVF counseling, clinical decision making, and outcomes.