NIPT time detection hierarchical algorithm based on multifactor clustering model

This study develops a computational framework integrating multilevel statistical modeling and machine learning to optimize testing scheduling for non-invasive prenatal testing (NIPT). Leveraging clinical NIPT datasets, a linear mixed-effects model was employed to quantify the significant effects of maternal body mass index (BMI), guanine-cytosine (GC) content, and sampling frequency on fetal Y-chromosome concentration. Specifically, each unit increase in maternal BMI was associated with a 0.0019-unit reduction in fetal Y-chromosome concentration. K-means clustering was utilized to stratify pregnant women into distinct risk groups, and the 25th percentile of gestational weeks was determined as the recommended testing time point—ensuring that no less than 75% of individuals in each group met the predefined detection threshold. Furthermore, a multi-factor predictive model identified a high-risk subgroup characterized by a mean BMI of 35.6, for which the recommended gestational week for NIPT was 18.3. This personalized testing scheduling strategy substantially enhances the reliability of NIPT and offers data-driven evidence to support clinical decision-making in prenatal care.