Introducing artificial intelligence and sperm epigenetics in the fertility clinic: a novel foundation for diagnostics and prediction modelling.

Abstract

Worldwide, infertility is a rising problem. A couple's lifestyle, age and environmental exposures can interfere with reproductive health. The scientific field tries to understand the various processes how male and female factors may affect fertility, but translation to the clinic is limited. I here emphasize potential reasons for failure in optimal treatment planning and especially why current prediction modelling falls short. First, Assisted Reproductive Technology (ART) has become a mainstream solution for couples experiencing infertility, while potential causes of infertility remain unexplored or undetermined. For instance, the role of men is generally left out of preconceptional testing and care. Second, regularly used statistical or computational methods to estimate pregnancy outcomes miss important biological and environmental factors, including features from the male side (e.g., age, smoking, obesity status, alcohol use and occupation), as well as genetic and epigenetic characteristics. I suggest using an integrated approach of biostatistics and machine learning methods to improve diagnostics and prediction modelling in the fertility clinic. The novelty of this concept includes the use of empirically collected information on the sperm epigenome combined with readily available data from medical records from both partners and lifestyle factors. As the reproductive field needs well-designed models at different levels, derivatives are needed. The objectives of patients, clinicians, and embryologists differ slightly, and mathematical models need to be adapted accordingly. A multidisciplinary approach where patients are seen by both, clinicians and biomedically skilled counsellors, could help provide evidence-based assistance to improve pregnancy success. Next, when it concerns factors that may change the ability to produce optimal embryos in ART, the embryologist would benefit from a personalized prediction model, including medical history of the patient as well as genetic and epigenetic data from easily accessible germ cells, such as sperm.