Combined Model-Based Prediction for Non-Invasive Prenatal Screening

Abstract

The risk of chromosomal abnormalities in the child increases with increasing maternal age. Although non-invasive prenatal testing (NIPT) is a safe and effective prenatal screening method, the accuracy of the test results needs to be improved owing to various testing conditions. We attempted to achieve a more accurate and robust prediction of chromosomal abnormalities by combining multiple methods. Here, three different methods, namely standard Z-score, normalized chromosome value, and within-sample reference bin, were used for 1698 reference and 109 test samples of whole-genome sequencing. The logistic regression model combining the three methods achieved a higher accuracy than any single method. In conclusion, the proposed method offers a promising approach for increasing the reliability of NIPT.