To apply this technique the researchers analyse initially potassium, urea and hypoxantine (a DNA metabolite) concentrations present in the vitreous humour of the eye of the human cadaver, and introduce these figures into a computer programme. The software that has been invented by these Galician scientists uses this information and is capable of establishing the time at which death occurred.
“The equations we have developed now make it possible for us to estimate the PMI more precisely than before, and provide a useful and accessible tool to forensic pathologists that is easy to use” José Ignacio Munoz Barús, one of the authors of the study, explains to SINC, and who is also a specialist doctor from the Institute of Legal Medicine at the University of Santiago de Compostela.
The traditional techniques for estimating the PMI are based on the study of parameters such as the rectal temperature of the cadaver or one of the organs, such as the liver, in rigor mortis, or post mortem lividity examination. These methods are complemented by biochemical analyses of the body fluids. One of these is the vitreous humour, the gelatinous liquid that is found behind the crystalline lens of the eye.
Muñoz Barús points out that the study, published recently in Statistics in
Medicine, suggests mathematical models that are “more flexible, useful and efficient” than those that have been applied until now. The doctor describes some of the previous techniques as “not very reproducible, not very precise and untested in the field”, such as the deterioration of DNA, immunoreaction or the traditional techniques based on the biochemistry of the vitreous humour.
In this last case the researcher specifies that previous studies used a “linear regression mathematical model” which assumes that the concentrations of potassium, hypoxantine and urea increase in a linear way that is more or less constant throughout the post mortem interval. However, the new analyses suggest that those premises are not valid and that the statistical models known as generalized additive models (GAM) or the support vector machine (SVM) models are more flexible and much more useful, since they avoid the assumption of linearity”.
The precision and usefulness of these two models have been confirmed by chemical analysis in more than 200 vitreous humour samples. The doctor and the two mathematicians who have performed the study have verified that the SVM method offers more precise data, although the GAM method is more easy to assimilate to the linear model and understand graphically and numerically, “ for which reason both complement each other”.
The three scientists have incorporated all this information into the development of a free computer package (based on code “R”) which makes it possible to establish the PMI using four predictive variables: concentrations of potassium, hypoxantine and urea, and cause of death. In addition, the software makes it possible to show the results graphically. “In this way the estimation of the time of death and expert examination are made easier when attending the courts of justice”, Munoz Barús points out to SINC
“The precise determination of the exact time of death has been the subject of various studies going back to the 19th century, since this information is of paramount importance in the field of legal medicine, owing to its repercussions on crime and civil society. This new method offers an important contribution to this field”, the researcher concludes.
Article: José Ignacio Muñoz Barus, Manuel Febrero-Bande, Carmen Cadarso-Suárez. “Flexible
regression models for estimating the postmortem interval (PMI) in forensic medicine”.
Statistics in Medicine 27 (24): 5026 – 5038, 2008.