Research on the Optimization of Pregnant Women's Characteristics and NIPT Detection Strategy Based on Data Driven Modeling

Abstract

This article focuses on key issues in non-invasive prenatal testing and establishes multiple regression/GAM, BMI clustering risk assessment, machine learning regression prediction, and classification models. By conducting relevant analysis to evaluate the relationship between gestational age, BMI, and Y chromosome concentration in pregnant women, multiple linear regression, second-order polynomial regression, and generalized additive model (GAM) were constructed. The results showed that the GAM model outperformed other models in R², better capturing the nonlinear effects of gestational age and BMI on Y concentration; The significance test showed that these factors had a statistically significant impact on Y concentration (p<0.05). Using K-means clustering to group pregnant women according to BMI, calculate the gestational weeks when the first Y concentration reaches 4% in each group, and construct a comprehensive risk function to consider the impact of the probability of reaching the standard and the detection time point on the risk, and solve for the optimal detection time point for each group. The results showed that there were significant differences in the first gestational age of different BMI groups, and the risk model recommended optimized detection time points for each group.This study provides a systematic, data-driven approach to optimize NIPT detection strategies by integrating regression analysis, clustering, and machine learning models. By revealing the nonlinear impact of gestational age and BMI on Y chromosome concentration and proposing differentiated detection time points for pregnant women with different BMI levels, the research not only improves the scientific basis for clinical decision-making but also reduces the risk of false negatives or delayed diagnoses. The findings contribute to more precise, personalized prenatal testing strategies, thereby enhancing the reliability and effectiveness of NIPT in practical medical applications.