Determination of gender based on osteometric fracture of the pelvic bones using discriminant analysis

Abstract

Backgraund: One of the current problems of forensic medicine is the sexual identification of a person based on bone remains. A separate difficulty is the conduct of sexual identification based on isolated bones. Thus, the study of the possibility of conducting sexual identification of a person based on an isolated pelvic bone is of great interest. .

Aims: The aim of the study was to develop a method for determining somatic sex based on osteometric features of skeletal pelvic bones using multivariate discriminant analysis. .

Materials and methods: The study is observational, retrospective, cross-sectional. The object of the study was skeletonized pelvic bones (n=42) of individuals of both sexes aged 18 to 86 years. from the collection of the Department of Human Anatomy of the Saratov State Medical University named after V.I. Razumovsky, as well as 25 pelvic bones from the collection of the Department of Human Anatomy, Operative Surgery and Topographic Anatomy of the Lugansk State Medical University. The measurements were carried out using a large thick compass, a sliding compass, and a centimeter tape according to the method of V. P. Alekseev. Classification models were built by step-by-step discriminant analysis.

Results: The study resulted in a model with a set of variables that allows determining the sex of bone remains based on osteometric parameters. The accuracy of the model identification was 90.4%. When tested on the verification group, the accuracy was 84%. Statistically significant differences were demonstrated by the following dimensions used in the mathematical model: pelvic bone height, ischial bone height, and the largest diameter of the acetabulum (p<0.01). .

Conclusions: The conducted research allowed to create a mathematical model that can be used for gender identification of bone remains based on 7 pelvic bone sizes; the theoretical accuracy of the model was 90.4%, and the practical accuracy was 84%.