Utilizing the Bayesian network algorithm for noninvasive prenatal paternity testing (NIPPT) and efficacy evaluation...

In recent years, noninvasive prenatal paternity testing (NIPPT) has gained substantial attention, and considerable progress has been made in its methodological development. However, most existing studies predominantly employ qualitative approaches for NIPPT data analysis and often overlook specific scenarios that may arise in practice, such as surrogacy or the absence of the alleged father. Furthermore, these studies typically involve only small sample sizes and lack a systematic evaluation of NIPPT efficacy. This raises concerns regarding the overall reliability of NIPPT results, which could impact their acceptance in legal contexts. In light of this background, the present study utilized the KinMix R package-developed within a Bayesian network framework-to conduct quantitative analysis of NIPPT data. We compare these results with those obtained from qualitative analysis to assess their influence on the weight of evidence. Additionally, we explored the feasibility of applying KinMix to other related forensic scenarios corresponding to different alternative hypotheses. Moreover, by generating simulated maternal-fetal mixed DNA data and analyzing them with KinMix, we established a method for assessing the efficacy of the NIPPT system. This study addresses current limitations in NIPPT research that have predominantly relied on qualitative analysis; it examines potential applications in specific scenarios while developing a systematic and scientific method for efficacy evaluation of NIPPT.