Abstract: Accurate age estimation from semen evidence is crucial for forensic investigations in sexual assault cases. While DNA methylation is a promising biomarker for predicting the donor's chronological age in forensic cases, most existing DNA methylation-based age estimation models primarily focus on somatic cells, with limited exploration of sperm-specific methylation signatures. Given that tissue-specific differences in CpG methylation may reduce the accuracy of existing epigenetic clocks for semen samples, there is a need to develop age-prediction models for this tissue in particular. For this study, we employed publicly available sperm methylation microarray datasets (GSE185920, n = 1471, aged 20-60 years) from the Gene Expression Omnibus (GEO) to identify age-related CpG sites (AR-CpGs). To identify AR-CpGs, we subsequently implemented a multi-algorithm feature selection strategy (maximum mutual information, L1 regularization, and sequential feature selection). We developed an optimized sperm-epigenetic clock by evaluating 69 machine learning regression model frameworks, achieving a mean absolute error (MAE) of 1.63 years in the training cohort. Validation on independent sperm datasets (GSE185445, n = 379, GSE149318, n = 90) yielded MAEs of 2.93 and 2.58 years, respectively, demonstrating robust generalization. To identify additional markers, we screened for sperm-specific AR-CpGs using whole-genome bisulfite sequencing (WGBS) data from the publicly available GEO dataset GSE222340. Subsequently, based on the pyrosequencing data of nine selected AR-CpG markers analyzed in 95 semen samples (ages 20-42 years), we developed a robust forensic model for human semen age estimation and determined the optimal algorithm by systematically evaluating 23 regression methods. The best-performing model, support vector machine (radial basis function kernel), exhibited an MAE of 2.21 years and a root mean square error (RMSE) of 3.15 years on the test set. This work provides a valuable set of AR-CpGs, develops an optimized sperm-chronological epigenetic clock, and delivers a practical model for estimating age from semen.