NIPT Testing Timing Decision Algorithm Based on K-means Clustering and Multi-Objective Risk Optimization

Abstract

The accuracy of non-invasive prenatal testing (NIPT) is highly dependent on fetal cell-free DNA concentrations, especially the 4% threshold for male Y chromosome concentrations. To optimize the timing of detection and minimize clinical risks, this paper proposes a data-driven decision-making framework. Firstly, the relationship between Y chromosome concentration and 21 characteristics such as gestational age and BMI was explored by Spearman rank correlation analysis and random forest regression model, and it was found that X chromosome concentration had the most significant effect. Secondly, a K-means clustering method based on standardized BMI was proposed for male fetal samples, and a comprehensive objective function combining early detection risk, failure risk, error risk and delay punishment was constructed to determine the optimal gestational age detection in each group. The results showed that the recommended testing time point for the low-to-medium BMI group (20.7-35.9) was about 14.9 weeks, while the high BMI group (36.3-45.7) needed to be delayed to 18.2 weeks, and the overall robust success rate was 0.849. Finally, the population was subdivided into four groups, with the optimal time point distributed from 14.1 to 18.5 weeks, and the overall robust success rate reached 0.863, of which the "standard compliance ratio" was identified as the most critical feature. This study provides a systematic algorithm solution for personalized NIPT detection time point planning.