Determination of Optimal Detection Time Window and Error Analysis for Male Fetal Non-Invasive Prenatal Testing Based on K-Means Clustering and Particle Swarm Optimization

Abstract

To address the issue that unreasonable timing of non-invasive prenatal testing (NIPT) increases detection risks, this study constructed a risk optimization model to determine the optimal NIPT time window for male fetuses. Taking gestational age and BMI as optimization variables, the study integrated the time-related risk from exponential fitting and the accuracy-related risk from density estimation, determined weights using the Analytic Hierarchy Process (AHP), and further established a total risk function. Pregnant women were divided into four groups by BMI via K-means clustering, and the Particle Swarm Optimization (PSO) algorithm was employed to solve the model. The results showed that: the optimal gestational ages corresponding to BMI ranges [26.6, 30.2), [30.2, 33.8), [33.8, 37.5), and [37.5, 46.9] were 11.85 weeks, 13.22 weeks, 15.78 weeks, and 19.33 weeks, respectively, indicating that the higher the BMI, the later the optimal detection time. Finally, an analysis of ±5% and ±10% concentration errors revealed that negative errors delay the optimal time point and reduce risks, while positive errors advance the time point and increase risks—with the high BMI groups being more significantly affected. This study provides a basis for personalized clinical testing.