A 2017 study by the World Health Organization (WHO) revealed that male infertility accounts for about half of all infertility cases. While semen analysis is crucial for diagnosing male infertility, it is often difficult to access, especially outside of fertility clinics. In light of this challenge, a team led by Associate Professor Hideyuki Kobayashi from Toho University School of Medicine in Tokyo has developed an innovative AI model that can predict the risk of male infertility using just a blood test to measure hormone levels. The model, which requires no programming skills to create, was detailed in the British scientific journal Scientific Reports.
The AI model, developed using data from 3,662 patients, showed an accuracy rate of about 74%. Remarkably, it predicted non-obstructive azoospermia, a severe form of male infertility, with 100% accuracy.
The study, conducted between 2011 and 2020, collected clinical data from men who underwent both semen and hormone tests for infertility. The semen tests measured semen volume, sperm concentration, and sperm motility, while the hormone tests looked at levels of LH, FSH, PRL, testosterone, and estradiol (E2). The total motile sperm count (TMSC), which combines these factors, was used to help assess male fertility. The AI model was trained using these variables and accurately predicted male infertility risk based on hormone levels alone.
Validation tests were conducted using data from 188 patients in 2021 and 166 patients in 2022. The model’s accuracy was 58% for the 2021 group and 68% for the 2022 group, but it maintained a perfect accuracy rate for detecting non-obstructive azoospermia in both years.
Associate Professor Kobayashi emphasized that the AI model should be used as an initial screening tool before more detailed semen testing. “While not a replacement for semen analysis, this model can be easily used in general medical facilities, making male infertility screening more accessible,” he explained. If the AI detects abnormal values, it should prompt patients to seek further testing at a specialized infertility clinic for appropriate treatment.
CreaTact, Inc., a Japanese software company, is working to develop a commercial version of this AI model for widespread use. Kobayashi hopes the model will help clinical laboratories and health checkup centers screen for male infertility, reducing the barriers to testing and improving accessibility.
This AI-powered approach offers hope for expanding access to male infertility testing, particularly in areas where specialized fertility clinics are scarce.
Related topics:
