TIMING SELECTION OF NIPT BASED ON A GENERALIZED LINEAR MIXED MODEL WITH A BINARY OUTCOME VARIABLE
Keywords:
K-means++ clustering, Two result variables, Combined stacking method, Generalized linear mixed modelAbstract
Non-invasive prenatal testing (NIPT) mitigates the uncertainties associated with childbirth and enhances neonatal quality, thereby facilitating policy implementation. Stratification by body mass index (BMI) can enhance the accuracy and safety of these tests. The study initially processed data and employed models to ascertain that maternal age, gestational age, BMI, and parity influence Y chromosome concentration. Subsequently, the K-means++ algorithm categorized pregnant women with male fetuses based on their BMI. Through survival analysis, the optimal NIPT timing for each group was predicted to be 12.9, 13.3, 13.9, 13.9, and 16.1 weeks. Monte Carlo simulations indicate that elevated testing errors disproportionately affect women with higher BMI. A novel model evaluates whether the Y chromosome meets established standards and assesses testing errors, yielding improved NIPT timings of 16.0, 17.3, 19.2, 21.1, and 12.0 weeks, thereby illustrating the impact of errors on timing. To detect abnormalities in female fetuses, a model utilizing random forest and light gradient boosting machine was developed. It achieved an area under the curve (AUC) of 0.9903 and an accuracy of 0.9889, delivering efficient and precise outcomes.References
[1] Budis J, Gazdarica J, Radvanszky J, et al. Non-invasive prenatal testing as a valuable source of population specific allelic frequencies. Journal of Biotechnology, 2019, 299: 26.
[2] Liu L L, Zheng M L, Zou B J, et al. Highly Sensitive and Specific Detection of SRY Gene for Non–invasive Prenatal Diagnosis. Chinese Journal of Analytical Chemistry, 2017, 45(10): 1043.
[3] Dasgupta S, Roy P, Ghosh C, et al. A prospective cohort study of adverse fetomaternal outcomes among overweight, obese, and normal weight pregnant women with term singleton pregnancy in a tertiary care hospital. Asian Journal of Medical Sciences, 2022, 13(4): 42351.
[4] Wang X H. The analysis of sculpture image classification in utilization of 3D reconstruction under K-means++. Scientific Reports, 2025, 15(1): 1949.
[5] Leivayamaguchi V, Alvares D. A Two-Stage Approach for Bayesian Joint Models of Longitudinal and Survival Data: Correcting Bias with Informative Prior. Entropy, 2020, 23(1): 50.
[6] Tang A M, Tang N S, Yu D L. Bayesian semiparametric joint model of multivariate longitudinal and survival data with dependent censoring. Lifetime Data Analysis, 2023, 29(4): 608.
[7] Abushattal M, Alhomaidat F, El Yabroudi M. Effects of intersection control types on driver yielding behavior to cyclists using mixed logit modeling. Scientific Reports, 2025, 15(1): 9801.
[8] Zhang J H, Xu Z J, Xiong Z, et al. Android malware detection based on feature fusion and the improved stacking ensemble model. Journal of Computer Virology and Hacking Techniques, 2025, 21(1): 546.
[9] Wang J J, Jiang L, Zhang L, et al. An Integrated Stacking Ensemble Model for Natural Gas Purchase Prediction Incorporating Multiple Features. Applied Sciences, 2025, 15(2): 778.
[10] Diamantopoulou M J, Özçelik R, Kalkanli Genç Ş. Evaluation of the random forest regression machine learning technique as an alternative to ecoregional based regression taper modelling. Computers and Electronics in Agriculture, 2025, 239(PA): 110964.
[11] Chen S Y, Shao W Q, Gao S P, et al. High impedance fault feeder selection method based on MPA-VMD and LightGBM in distribution networks. International Journal of Electrical Power and Energy Systems, 2025, 171: 110985.