Predicting Male Fertility Using Machine Learning: A Semen Parameters Based Analysis with the VISEM Dataset

Traditional semen analysis is often manual, subjective, and prone to inter-laboratory variability. This study investigates whether machine learning (ML) can provide a more objective, consistent, and accurate method for classifying male fertility status by analyzing standardized semen parameters—specifically sperm concentration, motility, and morphology—according to World Health Organization (WHO) criteria.

The researchers utilized the publicly available VISEM dataset, which contains semen analysis records from 85 male participants. They pre-processed the data and engineered features based on WHO thresholds to label each sample into one of three categories: Fertile, Sub-Fertile, or Infertile. To identify the most effective diagnostic tool, the team employed the LazyPredict framework to benchmark over 40 different supervised machine learning algorithms. The models were evaluated using 5-fold cross-validation, with performance metrics including accuracy, F1-score, and ROC-AUC analysis to ensure robustness.

Among the 40+ algorithms tested, the Nearest Centroid classifier emerged as the top performer, achieving an accuracy of 94.2%. This model outperformed other well-known techniques, such as Support Vector Machines (SVM) and Quadratic Discriminant Analysis (QDA). The study demonstrates that machine learning can effectively synthesize complex semen data into clear, clinically relevant fertility categories, offering a potential decision-support tool to reduce observer bias and improve the reliability of reproductive diagnostics.

By automating the classification of fertility status, this approach could standardize diagnostic workflows in andrology and assisted reproductive technologies (ART). Providing an objective, data-driven assessment helps clinicians make more consistent treatment recommendations, ultimately supporting better fertility planning for couples and reducing the diagnostic delays associated with manual, subjective semen evaluation.