Artificial intelligence in pregnancy prediction

Objective. To study the power of artificial intelligence in predicting the onset of pregnancy by evaluating patients’ hormonal status and medical history. Material and methods. A retrospective analysis was made of 334 case histories of patients aged 20 to 48 years who underwent in vitro fertilization (excluding cryoprocedures) between July 2011 and January 2019 using the SPERO-screen software algorithm. Results. The probability of pregnancy on dates significantly different from the software-calculated dates is significantly lower than the data reported in the world literature. Use of the data analysis algorithm created on the basis of artificial intelligence makes it possible to increase the probability of pregnancy by more than 2-fold. Conclusion. On the basis of the analysis of a woman’s hormonal status and anamnesis data with the application of machine learning technologies with the application of artificial neural networks, prepared on the basis of thousands of clinical cases of successful pregnancy, it is possible to predict the most favorable menstrual cycles and specific dates of conception.