OPTIMIZING THE TIMING OF NIPT BASED ON MULTI-FACTOR REGRESSION AND DECISION ANALYTICS

Abstract

Noninvasive prenatal testing (NIPT) is widely used for screening common fetal aneuploidies, but its reliability is strongly affected by fetal fraction and is often reduced in pregnant women with elevated body mass index (BMI). This study develops an integrated statistical and optimization framework to generate practical recommendations on testing timing under heterogeneous maternal characteristics and sequencing quality conditions. Using real-world NIPT records from male fetal samples, Y chromosome concentration was adopted as a proxy for fetal signal adequacy, and qualification was defined by the empirical threshold of Y ≥ 4%. First, multivariate regression and weighted least squares were used to quantify the joint effects of gestational age and BMI on Y chromosome concentration and to derive a closed-form qualification probability. Second, BMI-stratified logistic models were constructed to estimate the earliest gestational week required to achieve a target qualification probability, such as 90%, and to reveal systematic timing delays associated with increasing BMI. Third, a risk-based optimization strategy was applied to generate a smooth BMI-dependent timing trajectory by explicitly balancing non-qualification risk against delay cost, and robustness was assessed through perturbation-based resampling. Finally, a multi-factor logistic generalized linear model incorporating maternal age, weight, unique aligned reads, and GC content was established, and joint optimization of BMI boundaries and testing times was used to eliminate infeasible extreme-BMI subgroups and improve group-level fairness. The proposed framework translates observational associations into deployable timing recommendations and provides quantitative support for individualized NIPT timing decisions.