Yilmaz Riza, Erkaymaz Okan, Kara Erdogan, Ergen Kivanc
Department of Forensic Medicine, Medical Faculty, Bulent Ecevit University, Zonguldak, Turkey.
Department of Computer Engineering, Bulent Ecevit University, Zonguldak, Turkey.
J Forensic Sci. 2017 Mar;62(2):468-472. doi: 10.1111/1556-4029.13277. Epub 2016 Dec 1.
Fetal deaths are important cases for forensic medicine, as well as for criminal and civil law. From a legal perspective, the determination of whether a deceased infant was stillborn is a difficult process. Such a determination is generally made during autopsy; however, no standardized procedures for this determination exist. Therefore, new facilitative approaches are needed. In this study, three new support systems based on 10 autopsy parameters were tested for their ability to correctly determine whether deceased infants were alive or stillborn. Performances were analyzed and compared with one another. The artificial neural networks and radial basis function network models had 90% accuracy (the highest among the models tested), 100% sensitivity, and 83.3% specificity. Thus, the models presented here provide additional insights for future studies concerning infant autopsies.
死胎对于法医学以及刑法和民法来说都是重要案例。从法律角度来看，判定一名死婴是死产还是活产是一个艰难的过程。这种判定通常在尸检时进行；然而，目前尚无用于此判定的标准化程序。因此，需要新的辅助方法。在本研究中，基于10个尸检参数测试了三种新的支持系统，以确定其正确判定死婴是活产还是死产的能力。对各系统的性能进行了分析并相互比较。人工神经网络和径向基函数网络模型的准确率为90%（在所测试的模型中最高），灵敏度为100%，特异度为83.3%。因此，本文提出的模型为未来有关婴儿尸检的研究提供了更多见解。
