A Hybrid Modeling Approach for Optimizing Detection Timing Based on Male Fetal Y Chromosome Concentration and Individual Variability

Abstract

This study focuses on the correlation analysis and model construction of factors influencing male fetal Y chromosome concentration. First, prenatal data from female fetuses underwent preprocessing, including excluding samples outside the 10–25 week gestational range, handling samples with special gestational patterns, and converting gestational age to a uniform “week” unit. The Lilliefors test confirmed that variables such as age, gestational age, Y chromosome concentration, and BMI did not satisfy the assumption of normal distribution. At the population level, Spearman's rank correlation analysis revealed a weak positive correlation between Y chromosome concentration and gestational age, and a weak negative correlation with BMI, though neither reached statistical significance. Subsequently, a Support Vector Machine (SVM) model was constructed and trained to capture potential nonlinear relationships among variables, achieving 92.31% accuracy on the test set. Finally, individual-level analysis via binary linear fitting of randomly selected representative samples revealed that gestational age showed a significant positive correlation with Y chromosome concentration, while BMI effects varied among individuals. This confirms that population-level statistical patterns often fail to fully account for individual-specific characteristics. These findings underscore the critical importance of individual heterogeneity, challenging the conventional one-size-fits-all paradigm. This study provides a robust methodological foundation for advancing personalized NIPT strategies, moving beyond population averages towards precision prenatal care.