A Machine Learning-based NIPT Intelligent Timing Recommendation and Fetal Abnormality Assistance Diagnosis System

DOI: https://doi.org/10.62517/jmpe.202618308

Author(s)

Zihao Zhang, Simin Wang*, Junjie Jiang, Jia Sheng, Boyan Zhang, Zhongyi Li, Ziwu Wang

Affiliation(s)

Jiangsu Normal University KeWen College, Xuzhou, Jiangsu, China *Corresponding Author

Abstract

In response to the two major pain points in the clinical application of non-invasive prenatal genetic testing (NIPT): the “one-size-fits-all” approach to determining gestational age and the “single-indicator” method for diagnosing chromosomal abnormalities in female fetuses, this study proposes a machine learning-based NIPT intelligent timing recommendation and fetal anomaly auxiliary judgment system. First, using the K-Means unsupervised clustering algorithm and the elbow method to determine the optimal clustering number K = 3, the BMI of pregnant women is automatically categorized into low, medium, and high-risk groups, replacing the traditional fixed-range empirical categorization method. Second, based on local weighted regression (LOESS), curves are constructed for the attainment rates of Y-chromosome concentrations in each group, and the optimal NIPT testing times for each group are determined by the intersection points of horizontal and vertical cutoff lines, at 22.143 weeks, 24.286 weeks, and 27.714 weeks, respectively. Finally, seven core indicators, including BMI, chromosome Z value, GC content, sequencing read segment proportion, and X-chromosome Z value, are integrated into a random forest classification model to quantify the risk probability of chromosomal abnormalities in female fetuses. The AUC value achieved is 0.799. On this basis, K-Means clustering and the random forest model were integrated into a web system using Flask+Vue.js, creating a NIPT intelligent decision support system that covers the entire business process of data entry, model calculation, and result visualization. This provides a novel technological solution for personalized and precise decision-making in prenatal screening.

Keywords

Non-Invasive Prenatal Genetic Testing (NIPT); Machine Learning; K-Means Clustering; Random Forest; Intelligent Decision Support; Time-based Recommendations

References

[1]Lo YMD, Corbetta N, Chamberlain PF, et al. Presence of fetal DNA in maternal plasma and serum. The Lancet, 1997, 350(9076):485-487. [2]Canick JA, Palomaki GE, Kloza EM, et al. The impact of maternal plasma DNA fetal fraction on next generation sequencing tests for common fetal aneuploidies. Prenatal Diagnosis, 2013, 33(7):667-674. [3]Norton ME, Jacobsson B, Swamy GK, et al. Cell-free DNA analysis for noninvasive examination of trisomy. New England Journal of Medicine, 2015, 372(17):1589-1597. [4]Gil MM, Accurti V, Santacruz B, et al. Analysis of cell-free DNA in maternal blood in screening for aneuploidies: updated meta-analysis. Ultrasound in Obstetrics & Gynecology, 2017, 50(3):302-314. [5]Bian Xu-ming, Liu Jun-tao, Qi Qing-wei. Interpretation of the Expert Consensus on Prenatal Screening and Diagnosis of Maternal Peripheral Blood Fetal Free DNA. Chinese Journal of Obstetrics and Gynecology, 2021, 56(3):145-150. [6]Ashoor G, Syngelaki A, Poon LCY, et al. Fetal fraction in maternal plasma cell-free DNA at 11-13 weeks' gestation: relation to maternal and fetal characteristics. Ultrasound in Obstetrics & Gynecology, 2013, 41(1):26-32. [7]Wang E, Batey A, Struble C, et al. Gestational age and maternal weight effects on fetal cell-free DNA in maternal plasma. Prenatal Diagnosis, 2013, 33(7):662-666. [8]Shao Li, Liu Ting. Study on a Machine Learning Prediction Model for the Risk of NIPT Failure in Large-Scale Obese Pregnant Women. Journal of Practical Obstetrics and Gynecology, 2025, 41 (2):136-140. [9]Huang H. Large cohort analysis of maternal BMI affecting fetal cfDNA fraction and NIPT test failure rate. BMC Pregnancy Childbirth, 2023, 23(1):629.Breiman L. Random forests. Machine Learning, 2001, 45(1):5-32. [10]Bianchi DW, Chudova D, Sehnert AJ, et al. Noninvasive prenatal testing and incidental detection of occult maternal malignancies. JAMA, 2015, 314(2):162-169. [11]Kim SK, Hannum G, Geis J, et al. Determination of fetal DNA fraction from the plasma of pregnant women using sequence read counts. Prenatal Diagnosis, 2015, 35(8):810-815. [12]Breiman L. Random forests. Machine Learning, 2001, 45(1):5-32. [13]Liu Ruiruì, Zhao Yan, Li Hong. Progress in Research on Prenatal Screening Models Based on Machine Learning. Chinese Digital Medicine, 2022, 17(5):32-38. [14]Wang L, Li Y. Maternal BMI stratification via K-means and LOESS smoothing to determine personalized NIPT sampling gestational age. J Clin Lab Anal, 2025, 39 (7):e24987. [15]Zhang L, Mason S. Loess regression analysis of fetal Y chromosome concentration stratified by maternal BMI for NIPT timing optimization. Prenatal Diagnosis, 2026, 46(4):412-420. [16]Li Xue, Zhang Li, Zhao Xiaoyu. K-Means Clustering Combined with Hierarchical BMI and LOESS Curve Prediction for Optimal Blood Collection Gestational Age in NIPT. Chinese Maternal and Child Health Care, 2023, 38 (12):2268-2272. [17]Chen T, Guestrin C. XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD, 2016:785-794. [18]Hosmer DW, Lemeshow S, Sturdivant RX. Applied Logistic Regression (3rd ed.). Hoboken: John Wiley & Sons, 2013. [19]Liu Min, Chen Lu, Wang Yan. Optimization of the screening efficacy for sex chromosome abnormalities in female fetuses using a fusion of multi-sequencing features and random forest algorithms. Chinese Journal of Perinatal Medicine, 2024, 27 (5):361-366. [20]Zhou Jia, Wu Fan, Ma Xiaoyu. Building a Web System for Machine Learning-Assisted Decision-Making in NIPT Using Flask+Vue. Chinese Journal of Digital Medicine, 2024, 19 (3):78-82. [21]Lei Wenjia, Wu Qingqing. Progress in the Application of Interpretable Machine Learning in Individualized Screening of NIPT. Chinese Journal of Evidence-Based Medicine, 2026, 26 (1):102-108.