Image by Elena Έλενα Kontogianni Κοντογιάννη from Pixabay    

Male infertility is a significant issue affecting millions of couples worldwide. According to the World Health Organization (WHO), around 9% of couples globally face fertility problems, with male factors contributing to about half of these cases. Traditional methods for diagnosing male infertility typically start with semen analysis, which involves evaluating sperm motility, morphology, velocity, and concentration. This process is often carried out in specialized fertility clinics using microscopes and counting chambers by skilled embryologists. However, this method has its limitations, including the discomfort many men feel about providing semen samples and the variability in semen analysis results.

The Role of AI in Predicting Male Infertility

Recent advancements in artificial intelligence (AI) and machine learning have opened new avenues for predicting male infertility. Unlike traditional statistics, which focus on processing data collected for a specific purpose, machine learning starts with the data and searches for patterns and features within it. This approach is particularly advantageous when dealing with large datasets, such as those found in medical records.

In a recent study, researchers explored the potential of using AI to predict male infertility based solely on serum hormone levels, without the need for semen analysis. They collected data from 3,662 patients, including information on various types of infertility such as non-obstructive azoospermia (NOA), obstructive azoospermia (OA), cryptozoospermia, oligozoospermia, asthenozoospermia, and normal cases. The researchers extracted age, luteinizing hormone (LH), follicle-stimulating hormone (FSH), prolactin (PRL), testosterone, estradiol (E2), and the ratio of testosterone to estradiol (T/E2) from medical records to train their AI models.

How AI Models Work

The AI models used in the study were based on two platforms: Prediction One and AutoML Tables. Both models demonstrated high accuracy in predicting male infertility. The Prediction One-based AI model had an area under the curve (AUC) value of 74.4%, while the AutoML Tables-based model had an AUC receiver operating characteristic (AUC ROC) of 74% and an AUC precision-recall (AUC PR) of 77%. These metrics indicate the models' ability to balance sensitivity (true positive rate) and specificity (true negative rate), as well as their overall performance in identifying abnormal results.

The study revealed that FSH was the most significant hormone for predicting male infertility. Elevated FSH levels often indicate problems with sperm production, making it a crucial marker. The ratio of testosterone to estradiol (T/E2) was also a significant predictor. Men with a low T/E2 ratio often benefit from treatments that reduce E2 levels, improving their sperm quality. The AI model accurately predicted non-obstructive azoospermia (NOA), a condition where no sperm is present due to production issues, with a 100% match in both 2021 and 2022. This high level of accuracy suggests that AI could be a reliable tool for diagnosing male infertility.

Key Findings and Implications

The study found that the AI model could accurately predict non-obstructive azoospermia (NOA), a condition where no sperm is present due to production issues, with a 100% match in both 2021 and 2022. This high level of accuracy suggests that AI could be a reliable tool for diagnosing male infertility.

The AI-based screening method offers several advantages over traditional semen analysis. First, it only requires a blood sample, making it less invasive and more acceptable to men who might be uncomfortable with semen collection. Second, hormone levels in the blood are more consistent than semen parameters, leading to more reliable results. Finally, this method could serve as an alternative to home diagnostic kits, providing a convenient and accurate screening option.

Conclusion

Male infertility is a complex issue, but new AI-based screening methods offer a promising solution. By analyzing hormone levels in the blood, AI models can accurately predict male infertility, providing a less invasive and more consistent alternative to traditional semen analysis. As research continues, these AI methods could revolutionize the diagnosis and management of male infertility, helping more couples achieve their dream of starting a family.

In summary, the integration of AI in medical diagnostics, particularly in predicting male infertility, represents a significant advancement. Traditional methods, while effective, have limitations that can be addressed by AI's ability to analyze large datasets and identify patterns that may not be immediately apparent. The use of serum hormone levels as a predictive tool offers a less invasive, more consistent, and potentially more accessible method for diagnosing male infertility. As technology continues to evolve, the hope is that these AI-based methods will become more refined and widely adopted, ultimately improving outcomes for couples struggling with infertility.

Read Paper▕ Scientific Reports