Analysis of the Optimal Timing for NIPT and the Detection of Fetal Abnormalities Based on Cluster Analysis and Gradient Boosting Techniques

Mengtian Zhang; Jingjing Lv; Xiaodong Tian; Yingying Zhang

doi:10.62051/ijphmr.v6n2.07

Authors

Mengtian Zhang
Jingjing Lv
Xiaodong Tian
Yingying Zhang

DOI:

https://doi.org/10.62051/ijphmr.v6n2.07

Keywords:

Multi-model Optimization for Non-Invasive Prenatal Testing, Generalized Additive Model, Clustering and Grouping

Abstract

With the continuous growth of the world's population, ensuring the health of the fetus has become a top priority. To optimize the timing selection of Non-Invasive Prenatal Testing (NIPT) and the strategy for determining fetal abnormalities, this paper constructs multiple models based on medical clinical knowledge and solves for the best NIPT under different circumstances. To explore whether the BMI of pregnant women affects the accuracy of NIPT, the correlation between the concentration of fetal Y chromosome, gestational age of pregnant women, and BMI is first analyzed visually through a correlation heatmap. Then, a generalized additive model is constructed, including single models of BMI, gestational age, and Y chromosome concentration, as well as an interaction model of BMI, gestational age, and Y chromosome concentration to obtain the correlation model. The model is solved using algorithms such as PIRLS. Through analysis, it is found that the concentration of Y chromosome is positively correlated with gestational age and weakly negatively correlated with BMI; both gestational age and BMI have statistical significance with Y chromosome concentration. To further focus on the BMI grouping of pregnant women carrying male fetuses and analyze the impact of different BMIs on the time to reach the Y chromosome concentration standard (>4%), avoiding subjective grouping errors and achieving the goal of "minimizing potential risks", the clustering analysis method is used to divide the BMI intervals with significant differences in the time to reach the Y chromosome concentration standard. Through the analysis of the interference of detection errors on the determination of the standard time, the best NIPT time points for each group are matched. The BMI grouping strategy based on clustering analysis and the matched best NIPT time points can significantly improve the detection accuracy, reduce the potential risks of pregnancy, and keep the detection errors controllable, providing reliable support for the precise clinical implementation of NIPT.

References

[1] Xue Ying, Zhao Guodong, Qiao Longwei, et al. Research Progress of Fetal Cell-Free DNA Enrichment Technologies in Maternal Peripheral Blood for Non-Invasive Prenatal Testing (NIPT) [J]. Chinese Journal of Birth Health & Heredity, 2023, 31(5): 1087-1090.

[2] Owen A. "Do you want to know or not?" How prenatal providers manage clinical uncertainty related to chromosomal risk and noninvasive prenatal testing [J]. Health (London), 2025, Epub ahead of print: 13634593251377111.

[3] Wang Weipeng, Wang Zhiguo, Zou Lin. Laboratory Management of Neonatal Disease Screening and Prenatal Diagnosis [M]. Beijing: People's Medical Publishing House, 2018.

[4] Li Chunjing, Wang Yue, Meng Yelin, et al. Smoothing Estimation Method for Functional Additive Quantile Regression Model [J]. Journal of Jilin Normal University (Natural Science Edition), 2025, 46(3): 56-67.

[5] Wu Chuanxu. Application of Nonlinear Regression Models [J]. Journal of Shaanxi University of Technology, 1997, (1): 60-62+66.

[6] Xiao W, Mao K, Liu H. Generalized Partially Functional Linear Model with Interaction between Functional Predictors [J]. Axioms, 2024, 13(9):583-583.

[7] Friedman J.H. Greedy Function Approximation: A Gradient Boosting Machine [J]. The Annals of Statistics, 2001, 29(5): 1189-1232.

[8] Giordani P, Kiers L A H. Weighted least squares for archetypal analysis with missing data [J]. Behaviormetrika, 2024, 51(1):441-475.

[9] Liu P, Yuan H, Ning Y, et al. A modified and weighted Gower distance-based clustering analysis for mixed type data: a simulation and empirical analyses [J]. BMC medical research methodology, 2024, 24 (1): 305.

[10] Pan Y, Pan X, Zhuang D, et al. A statistical investigation of parameters associated with low cell-free fetal DNA fraction in maternal plasma for noninvasive prenatal testing [J]. The journal of maternal-fetal & neonatal medicine : the official journal of the European Association of Perinatal Medicine, the Federation of Asia and Oceania Perinatal Societies, the International Society of Perinatal Obstetricians, 2024, 37(1):2338440-2338440.