Application of discriminant function analysis and logistic regression models to estimate sex using the dimensions around the metatarsal diaphyseal nutrient foramen in the South Africans of Mixed ancestry
Arthur Tsalani Manjatika Joshua Gabriel Davimes Pedzisai Mazengenya a School of Anatomical Sciences,University of the Witwatersrand,Johannesburg,South Africab School of Life Sciences and Allied Health Professions,Anatomy Division,Kamuzu University of Health Sciences,Blantyre,Malawic College of Medicine,Ajman University,Ajman,United Arab Emiratesd Center of Medical and Bio-Allied Health Sciences Research,Ajman University,Ajman,United Arab Emiratese Faculty of Health Sciences,University of the Witwatersrand,Johannesburg,South Africa
DOI: https://doi.org/10.1080/00450618.2024.2394419
2024-08-28
Australian Journal of Forensic Sciences
Abstract:The development of sex estimation standards, especially in locations with high murder case loads like South Africa, is important to aid in establishing the biological profile of victims. There is a lack of information on the utility of the measurements around the metatarsal diaphyseal nutrient foramina (NF) in sex estimation in South Africans of Mixed ancestry (SAMA). Five measurements around the NF were taken from a total of 248 metatarsals (first to fifth) from 28 males and 23 females of SAMA population. Measurements subjected to direct and stepwise discriminant function (DFA) and logistic regression (LRA) analyses included total length, distance from proximal end to NF, circumference, and mediolateral and dorsoplantar diameters at the level of the NF. The univariable functions produced 60.8–80% accuracy rates. The original classification accuracies for the stepwise and direct DFA multivariable functions ranged from 75.5–80% and 75–81.3%, respectively. The original classification accuracies for the stepwise and direct LRA multivariable functions ranged from 75.5%–80% and 75%–83.7%, respectively. The cross-validation classifications showed a drop of 0–4% for DFA and 0.7–5.4% for LRA. The multivariable DFA and LRA functions produced high average classification accuracies appropriate for use in forensic settings.
medicine, legal