Author(s)

DaZhi Wei

Affiliation(s)

College of Electronic Information and Automation, Tianjin University of Science and Technology, Tianjin 300222, China.

Corresponding Author

DaZhi Wei

ABSTRACT

To address the challenge of detecting chromosomal abnormalities in female fetuses due to the absence of Y chromosome data in non-invasive prenatal testing (NIPT), this paper proposes an innovative dual-layer classification model that integrates logistic regression and random forest. The model comprehensively utilizes 16-dimensional features including Z-scores and GC content of chromosomes 13, 18, and 21, along with key maternal clinical indicators. Through rigorous statistical testing and feature importance analysis, seven key discriminatory features were identified, establishing a progressive "abnormality screening-disease typing" judgment process. The framework employs an ensemble approach where logistic regression provides interpretable initial screening while random forest handles complex non-linear patterns for fine-grained classification. After cross-validation and threshold optimization, the model ultimately achieved an impressive accuracy rate of 99.57%, with precision and recall rates exceeding 98.5% across all abnormality categories. Comparative experiments demonstrated the superiority of this hybrid approach over single-model methods, particularly in handling imbalanced data distributions. The core innovation of this research lies in the integration of feature fusion and model collaboration, enabling high-precision, automated detection of chromosomal abnormalities in female fetuses and providing a new technical pathway for clinical precision diagnosis.

KEYWORDS

Non-invasive prenatal testing (NIPT); Chromosomal abnormalities in female fetuses; Dual-layer classification model; Feature selection; Random forest

CITE THIS PAPER

DaZhi Wei. Detection of chromosomal abnormalities in female fetuses based on a fused logistic regression-random forest model. World Journal of Information Technology. 2025, 3(6): 22-28. DOI: https://doi.org/10.61784/wjit3072.

REFERENCES

[1] Lo Y M D. Non-invasive prenatal testing by next generation sequencing: maternal plasma DNA and RNA. Annual Review of Genomics and Human Genetics, 2022, 23, 413-431.

[2] Bianchi D W, Chiu R W K. Sequencing of circulating cell-free DNA during pregnancy. New England Journal of Medicine, 2022, 379(5): 464-473.

[3] Norwitz E R, Levy B. Noninvasive prenatal testing: the future is now. Reviews in Obstetrics and Gynecology, 2023, 12(2): 89-95.

[4] Zhang Yan, Li Qiang, Liu Shuzheng, et al. Classification of chromosomal abnormalities in noninvasive prenatal testing based on machine learning algorithms. Bioinformatics, 2020, 18(3): 156-162.

[5] Chen Si, Liu Pei, Zhao Yang, et al. Clinical application of whole genome sequencing-based noninvasive prenatal testing in 20,000 pregnancies. Chinese Journal of Obstetrics and Gynecology, 2022, 57(2): 89-95.

[6] Wang Ke, Li Hui, Yuan Ming, et al. Detection of fetal aneuploidy by dual-model algorithm based on maternal plasma DNA sequencing. Chinese Journal of Medical Genetics, 2019, 36(5): 412-418.

[7] Gregg A R, Skotko B G, Benkendorf J L, et al. Noninvasive prenatal screening for fetal aneuploidy, 2016 update: a position statement of the American College of Medical Genetics and Genomics. Genetics in Medicine, 2016, 18(10): 1056-1065.

[8] Huang Rong, Li Ming, Wang Shu, et al. Comparative study of machine learning methods for feature selection and classification in noninvasive prenatal testing. Chinese Journal of Biomedical Engineering, 2020, 39(2): 156-163.

[9] Xu Jing, Chen Liang, Wang Rui, et al. A deep learning framework for fetal chromosomal abnormality detection from low-coverage sequencing data. Chinese Journal of Perinatal Medicine, 2022, 25(1): 45-51.

[10] Wu Qian, Zhou Ying, Li Xue, et al. Clinical validation of a random forest-based noninvasive prenatal testing model in 15,456 pregnancies. Chinese Journal of Obstetrics & Gynecology and Pediatrics, 2021, 17(3): 289-295.

[11] Petersen A K, Cheung S W, Smith J L, et al. Positive predictive value estimates for cell-free noninvasive prenatal screening from data of a large referral genetic diagnostic laboratory. American Journal of Obstetrics and Gynecology, 2017, 217(6): 691.e1-691.e6.

[12] Liang Xue, Wang Tao, Chen Yang, et al. A cost-effective method for noninvasive prenatal screening using low-pass whole genome sequencing. Chinese Journal of Practical Gynecology and Obstetrics, 2020, 36(8): 712-718.