欢迎访问新加坡聚知刊出版有限公司官方网站
info@juzhikan.asia
Integrating GAM and Survival Analysis: A Multi-Model Collaborative Framework for Optimizing NIPT Timing and Chromosomal Abnormality Determination in High-BMI Pregnant Women
  • ISSN:3041-0843(Online) 3041-0797(Print)
  • DOI:10.69979/3041-0843.25.04.084
  • 出版频率:Quarterly Publication
  • 语言:English
  • 收录数据库:ISSN:https://portal.issn.org/ 中国知网:https://scholar.cnki.net/journal/search

Integrating GAM and Survival Analysis: A Multi-Model Collaborative Framework for Optimizing NIPT Timing and Chromosomal Abnormality Determination in High-BMI Pregnant Women
Yu Yangyi  Zhao Zhenyi  Yi Yaxuan  Zheng Shican

School of Materials Science and EngineeringShenyang Aerospace UniversityShenyang110136;

Abstract:The application of non-invasive prenatal testing (NIPT) in pregnant women with high body mass index (BMI) faces challenges such as significant fluctuations in fetal cell-free DNA concentration, inconsistent testing timepoints, and complex chromosomal abnormality determination. Based on NIPT data from 1081 high-BMI pregnant women, this study constructed a collaborative analysis framework integrating generalized additive models (GAM), survival analysis, and multi-model fusion classification. First, GAM revealed nonlinear relationships between Y chromosome concentration and gestational age/BMI, achieving 82% model explanatory power. Second, Kaplan-Meier curves and Cox proportional hazards models stratified male fetus pregnancies by BMI and multifactorial risk factors. This identified optimal testing windows across groups that balanced high expected pass rates (≥90%) with low potential risks, while validating window stability. For female fetal chromosomal anomaly detection, a multi-output classification model integrating RandomForest, XGBoost, and LightGBM demonstrated superior performance in detecting abnormalities on chromosomes 13, 18, and 21 (most metrics exceeding 0.98). SHAP analysis further confirmed that the Z-score of the target chromosome serves as the core predictive feature. This study provides a comprehensive, interpretable clinical decision support solution for high-BMI pregnant women, spanning concentration prediction, timing optimization, and anomaly determination.

Keywords: Generalized additive model; Kaplan-Meier survival analysis; Cox proportional hazards model; LightGBM multi-output classification; SHAP interpretability analysis; Non-invasive prenatal testing

References

[1] Application of NIPT in prenatal screening for fetal chromosomal aneuploidy. Liu Li; Zhao Jinsong; Wang Dan; Xu Yan; Qi Chuan; You Junling; Yu Xiaochuan; Zhou Li. Chinese Journal of Eugenics and Genetics, 2019(11).

[2] Clinical Significance of NIPT in Detecting Rare Autosomal Aneuploidies. Hu Liang; Liu Jinxing; Wen Lijuan; Pei Yuanyuan; Luo Qi; Wei Fengxiang. Chinese Journal of Prenatal Diagnosis (Electronic Edition), 2022(03).

[3] Issues Related to NIPT for Prenatal Screening of Common Fetal Chromosomal Aneuploidies. Liu Juntao. Journal of Practical Obstetrics and Gynecology, 2018(11).

[4] Application of NIPT in Prenatal Screening for Fetal Chromosomal Aneuploidy. Zhi Diyuan. Chinese Journal of Occupational Health, 2021(05).

[5] Research Progress on NIPT Application in Diagnosing Fetal Aneuploidy Malformations. Teng Yue; Liu Dan; Huang Pu; Li Xu; Li Chunfang. Chinese Journal of Maternal and Child Health Research, 2016(02).