Author(s)

Jie Qian, Xuanyu Lu

Corresponding Author

Jie Qian

Abstract

With the rapid development of medical technology, NIPT, a non-invasive prenatal testing technology, has become one of the important tools for early detection and determination of fetal health. Based on the K-means clustering and greedy algorithm, this study focuses on the selection of the optimal NIPT time point for different pregnant women. Firstly, the three-dimensional clustering of Y chromosome concentration, gestational age and BMI was grouped by K-means clustering, and a cubic polynomial relationship model between Y chromosome concentration and gestational age and BMI was established. Then, for the problem of BMI grouping of male fetuses and pregnant women, a single-objective optimization model was established and the greedy algorithm was used to solve the best detection gestational age of each BMI grouping, and the results showed that different BMI intervals corresponded to different optimal detection times, and the detection error had an impact of 6.09% on the results. Finally, considering multiple factors such as height, weight, and age, a multiple nonlinear regression model was established and the detection time point was optimized, and the detection error influence was 16.16%. The results show that the optimization model using clustering and greedy algorithms can effectively determine the time point of personalized NIPT detection, providing a theoretical basis for clinical practice.

Keywords

NIPT; K-Means Clustering; Greedy Algorithm; Multiple Linear Regression