Chemosteometric regression models of heat exposed human bones to determine their pre-burnt metric dimensions

Heat exposure can lead to apparently random osteometric changes that hinder the application of metric methods used for biological profiling. The impracticality of using objective and burn-specific osteometric methods reduces the chances of establishing the biological profiles of unknown individuals based on their skeletal remains. We investigated the potential of chemometry analysis based on infrared spectroscopy to predict the amount of heat-induced osteometric changes and how this reflected into sex estimation.

Bones from 41 identified adult skeletons (24 females and 17 males with ages between 62 and 90 years old) were experimentally burnt to maximum temperatures ranging from 450°C to 1,100°C (attained after 65 to 240 min). Measurements were taken both before and after each experiment and powder samples were analyzed through FTIR-ATR. Correlations among heat-induced metric changes and chemometric indices (crystallinity index; B-type carbonates; carbonate [A + B] to carbonate B ratio; hydroxyl to phosphate ratio; 630 cm^-1 , 1450 cm^-1 , 3572 cm^-1 , and 3642 cm^-1 ) were tested. Significant variables were used to build regression models to predict heat-induced metric change which were then tested on an independent set of samples. Agreement in sex estimation between the pre- and post-burnt samples was also evaluated.

All indices were significantly correlated to heat-induced metric changes (α = .01) and the highest correlations were obtained for the 630 cm^-1 , 3572 cm^-1 , and crystallinity index. We confirmed that regression models based on chemometrics obtained from infrared spectra through FTIR-ATR are better at estimating heat-induced metric changes affecting bone and at sexing remains than other osteometric methods such as those based on correction factors or on metric references specific to calcined bones.

Regression models avoid the subjectivity associated with the application of other methods. While the latter can be applied only to calcined bones, which is difficult to assess sometimes, regression models can be applied to all bones regardless of their condition. Also, regression models have the advantage of allowing to infer about heat-induced metric change on a case-by-case basis.