Incorporating A.I. [Artificial Intelligence] into Boar Semen A.I. [Artificial Insemination

Abstract

Analysis of the U.S swine herd shows variation in pregnancy rate is more attributable to male- factor subfertility than the dam. To date, a limited degree of correlations has been observed between conventional semen analysis parameters and actual fertility after standard quality cutoffs are met. Thus, a clear ability to predict male-factor fertility is lacking. Recent technological advances in flow cytometry, namely image-based flow cytometry, allows for high-throughput, single-cell phenotyping. This allows for multi-million sperm bioimage data sets to be easily attained. These datasets can then be analyzed utilizing machine & deep learning analytic methods and correlated with single sire field fertility data to open the black box of boar fertility prediction. In this presentation, we will discuss recent advancements our lab has made utilizing machine & deep learning, artificial intelligence into boar sperm quality analysis. This includes bioimage, deep learning analysis of boar sperm morphology as well as our latest results related to boar semen fertility prediction. Our findings establish a new era for boar sperm quality analysis and pave the way for accurate fertility prediction in future precision agriculture applications. Work performed here was supported by USDA’s National Institute of Food and Agriculture grant number 2019- 67012-29714 (KK) and 2022-67015-36298 (KK).