Department of Anatomy, Midwestern University, Downers Grove, IL, USA.
Division of Forensic Medical Examination, National Forensic Service, 26460 Wonju, South Korea.
Forensic Sci Int. 2022 Jul;336:111318. doi: 10.1016/j.forsciint.2022.111318. Epub 2022 Apr 28.
This study addresses issues associated with limited access to reference skeletal collections and how it hampers regional methodological development. The study also investigates the effects of three different priors on age-at-death estimations of Koreans and quantifies the accuracy of Bayesian estimation compared to a non-Bayesian approach. CT scan images of pubic symphyses were scored following the Suchey-Brooks standards. We specifically examine Bayesian age-at-death estimations under an optimal scenario where both the likelihood and prior probability are derived from a target population of Korean origin. Three different priors (population, forensic, and uniform priors) are evaluated for their age estimation accuracy levels. The results showed that Korean males and females had discrepancies in skeletal aging rates only in certain phases. A similar trend was observed in the comparison of the original Suchey-Brooks and Korean sample aging rates. However, when final age estimates were computed based on an inadequate choice of a prior, the between-group differences may be minimized or even obscured. The uniform and forensic priors produced the most reasonable and unbiased estimates, while the uniform prior consistently provided open-ended estimates for the last phase. The population prior produced consistently higher estimates due to the high survivorship of Koreans. In addition, the original Suchey-Brooks standards persistently under-estimated ages for Korean males and females. Overall, when Bayesian age estimation is based on an informed prior that best matches the context of the case (forensic deaths in this study), there was increased accuracy of final age estimates compared to age estimates of other priors and the original Suchey-Brooks method.
本研究探讨了与有限获取参考骨骼标本相关的问题，以及这些问题如何阻碍区域性方法的发展。该研究还调查了三种不同先验概率对韩国人年龄推断的影响，并与非贝叶斯方法相比，量化了贝叶斯估计的准确性。对耻骨联合的 CT 扫描图像进行了评分，评分标准采用 Suchey-Brooks 标准。我们专门在最理想的情况下，即似然和先验概率均来自韩国人群的情况下，研究了贝叶斯年龄推断。我们评估了三种不同的先验概率（总体、法医和均匀先验概率）的年龄估计准确性水平。结果表明，韩国男性和女性的骨骼老化率仅在某些阶段存在差异。在比较原始的 Suchey-Brooks 年龄推断和韩国样本年龄推断时，也观察到了类似的趋势。然而，当根据不合适的先验概率计算最终年龄估计时，组间差异可能会最小化甚至被掩盖。均匀和法医先验概率产生了最合理和无偏的估计，而均匀先验概率在最后一个阶段始终提供无限制的估计。由于韩国人的存活率较高，总体先验概率始终产生较高的估计。此外，原始的 Suchey-Brooks 标准持续低估韩国男性和女性的年龄。总的来说，当贝叶斯年龄估计基于最符合案例背景的信息先验（本研究中的法医死亡）时，与其他先验概率和原始 Suchey-Brooks 方法相比，最终年龄估计的准确性有所提高。
