Population-inclusive assigned-sex-at-birth estimation from skull computed tomography scans

OA Version
Citation
S.R. Kelley, S.D. Tallman. 2022. "Population-Inclusive Assigned-Sex-at-Birth Estimation from Skull Computed Tomography Scans" Forensic Sciences, Volume 2, Issue 2, pp.321-348. https://doi.org/10.3390/forensicsci2020024
Abstract
Methods for estimating assigned, binary sex at birth from skeletonized remains have primarily been developed for specific population groups in the U.S. (e.g., African American, European American, Hispanic) and, thus, inherently rely on ancestry estimation as a foundational component for constructing the biological profile. However, ongoing discussions in forensic anthropology highlight pressing issues with ancestry estimation practices. Therefore, this research provides population-inclusive assigned-sex estimation models for cases where ancestry is not estimated or is truly unknown. The study sample (n = 431) includes 3D volume-rendered skull computed tomography scans from the novel New Mexico Decedent Image Database of African, Asian, European, Latin, and Native Americans. Five standard nonmetric traits were scored, and eighteen standard measurements were obtained. Binary logistic regressions and discriminant function analyses were employed to produce models and classification accuracies, and intraobserver reliability was assessed. The population-inclusive nonmetric and metric models produced cross-validated classification accuracies of 81.0–87.0% and 86.7–87.0%, respectively, which did not differ significantly from the accuracy of most population-specific models. Moreover, combined nonmetric and metric models increased accuracy to 88.8–91.6%. This study indicates that population-inclusive assigned-sex estimation models can be used instead of population-specific models in cases where ancestry is intentionally not estimated, given current concerns with ancestry estimation.
Description
License
Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).