A groundbreaking study led by Associate Professor Hideyuki Kobayashi at Toho University School of Medicine in Tokyo, Japan, has introduced an innovative AI model capable of predicting male infertility risk without the traditional need for semen analysis. Published in Scientific Reports, the research leverages artificial intelligence created without traditional programming, marking a significant advancement in diagnostic technology.
Traditionally, semen analysis has been pivotal in diagnosing male infertility, but accessibility outside specialized infertility clinics has been limited. The AI model developed by Kobayashi’s team uses data from 3,662 patients and achieves an impressive accuracy rate of approximately 74%. Notably, it boasts a perfect 100% accuracy in identifying non-obstructive azoospermia, the severest form of male infertility.
The study, conducted between 2011 and 2020, collected clinical data encompassing semen and hormone tests. Parameters such as semen volume, sperm concentration, motility, and hormone levels including LH, FSH, PRL, testosterone, and E2 were analyzed. The AI model derives a Total Motile Sperm Count (TMSC) from these tests, crucial for predicting infertility risk.
Validating the model with data from subsequent years (2021 and 2022), the accuracy varied but remained promising. In 2021, with 188 patients, accuracy reached approximately 58%, and in 2022, with 166 patients, it improved to around 68%. Importantly, the AI maintained its flawless prediction of non-obstructive azoospermia, confirming its reliability in severe cases.
Associate Professor Kobayashi emphasized that while the AI model serves as an initial screening tool, it does not replace traditional semen analysis. Instead, it offers a preliminary assessment that can be conveniently conducted at various medical facilities, facilitating early detection and subsequent referral to infertility specialists when abnormalities are detected.
Looking ahead, CreaTact, Inc. is advancing software development for a commercial AI prediction model, aiming to enhance accessibility to male infertility screening in clinical and health checkup settings. Associate Professor Kobayashi expressed optimism that widespread adoption of their AI model will democratize access to male infertility testing, addressing current barriers in healthcare.
Related Links:
