The application of a CART model for forensic human geolocation using stable hydrogen and oxygen isotopes

Abstract

AbstractThe utility of stable hydrogen and oxygen isotope analysis of human tissues for geolocation is an important area of study within forensic science. This study aimed to first validate the latitudinal relationship between stable hydrogen and oxygen isotopes in drinking water and human keratinous tissues through the analysis of human samples with known geographical origin. And secondly, to explore the use of classification and regression tree (CART) models to geographically classify individuals based on the stable isotope values of tissues themselves. Human hair and toenails were collected from four distinct study sites across Canada. The comparison of stable isotope values in drinking water and human tissues produced low R2 values indicating that linear models may not fully explain the variation observed for both hydrogen and oxygen values. Additionally, large intrapopulation variations were observed for Canadian cities and highlights the importance of understanding the regional isotopic spread of human values. Further, this study demonstrated that a closed group of unknown individuals known to have originated from a limited number of geographically distinct regions may be classified into their respective groups through the use of CART models. The potential for the CART model approach for human geolocation presents a promising new tool.