A recent study published in Scientific Reports examines the efficacy of a novel artificial intelligence (AI) screening method that assesses male infertility through serum hormone levels. Infertility is a significant global issue, impacting about 9% of the population, which translates to approximately 72.4 million individuals. Among these cases, male infertility accounts for 50% of infertility diagnoses, typically evaluated through semen analysis and hormonal assessments.

Importance of Semen Analysis

Semen analysis is crucial in understanding sperm production and maturation within the testes, as well as evaluating the health of the seminal pathways and the secretory activity of testicular glandular cells. The World Health Organization (WHO) provides a Laboratory Manual for the Examination and Processing of Human Semen, which outlines the standards for analyzing various semen parameters.

However, traditional semen analysis methods come with several limitations. Sample collection can be a significant barrier, as many men may hesitate to undergo testing due to social stigma. Moreover, the manual inspection of sperm is labor-intensive and requires skilled personnel, highlighting the urgent need for alternative screening methods for male infertility.

Hormonal Factors in Sperm Production

Normal sperm production relies on both testicular and endocrine functions, initiated by the hypothalamo-pituitary-testicular axis. Key hormones involved in this process include:

• Luteinizing Hormone (LH)
• Follicle-Stimulating Hormone (FSH)
• Prolactin (PRL)
• Testosterone
• Estradiol (E2)

FSH and LH are released from the anterior pituitary gland in response to gonadotropin-releasing hormone (GnRH) from the hypothalamus. FSH is essential for initiating spermatogenesis through its action on Sertoli cells within the testes. Inhibin B and testosterone, secreted by Sertoli and Leydig cells respectively, play significant roles in regulating the reproductive axis, with both hormones influencing further hypothalamic and pituitary activity.

In cases of spermatogenesis disorders, elevated FSH levels may occur without corresponding changes in LH and testosterone. Therefore, understanding the relationship between serum hormone levels and semen analysis is critical.

Study Overview

The recent study involved 3,662 patients who had previously undergone semen analysis and hormone level measurements as part of their male infertility evaluation. The average age of participants was 36 years. Among the patients, approximately 44% presented with oligozoospermia or asthenozoospermia, indicating reduced sperm count and motility, respectively. Azoospermia, characterized by a total absence of sperm, affected 12.2% of the cohort, with 5.7% classified as obstructive azoospermia (OA) and the remainder as non-obstructive azoospermia (NOA).

The research team utilized various hormonal levels, including LH, FSH, PRL, E2, and the ratio of testosterone to estradiol (T/E2), to train their AI models and assess the likelihood of male infertility. The lower limit for normal sperm levels was defined as a total motility sperm count of 9.408 × 10^6.

The AI model developed by Prediction One yielded an area under the curve (AUC) value of 74.4%, indicating its predictive performance. Using AutoML Tables, the AI model achieved an AUC receiver operating characteristic (ROC) of 74% and an AUC precision-recall (PR) of 77%. A higher AUC reflects better model performance, with the ROC curve measuring sensitivity versus specificity and the PR curve balancing accuracy with comprehensive identification of abnormal results.

In this study, FSH emerged as the most significant hormone in predicting male infertility, followed by the T/E2 ratio and LH. The AI model demonstrated a perfect match for diagnosing NOA and male hypogonadism, with a 70% accuracy for OA cases.

Conclusions and Future Implications

Previous research has suggested the potential for AI to predict endocrine statuses. Applications have included predicting postoperative outcomes following pituitary surgery or evaluating elevated parathyroid hormone (PTH) levels.

The AI models developed in this study have demonstrated high accuracy in predicting male infertility. FSH’s predictive significance is underscored by its role in spermatogenesis, while the importance of the T/E2 ratio reflects historical treatments with aromatase inhibitors in patients with elevated ratios.

The researchers propose their AI model as a convenient screening tool for male infertility before undergoing traditional semen analysis. Given that infertile men often have higher rates of medical comorbidities, this screening approach could not only assess infertility risk but also provide insights into overall health. While this AI model is not intended to replace semen analysis, it offers a promising alternative to conventional diagnostic kits.