Baylor University researchers have found a promising new method to determine the date of skeletal remains.
The relatively simple technique of applying statistics to chemical measurements could provide a quicker way for crime scene investigators and others to determine the post-mortem interval of bones, or the time that has elapsed since a person has died.
It is believed to be the first time that chemometric modeling of spectral data has been used to determine the time elapse after death of skeletal remains. In laboratory tests using this method, Baylor researchers found an error rate of only four to nine days for bones that were up to 90-days old.
"In perfect conditions in the laboratory, the method looks very encouraging," said Dr. Kenneth Busch, professor of chemistry and co-director of the Center for Analytical Spectroscopy at Baylor and a lead investigator on the project. "Once a regression model is built from spectral data, you could find out the age of the bones in a matter of minutes, rather than taking hours or days."
Once skeletization occurs, few techniques exist to determine the post-mortem interval of remains in a timely matter. In regions with high heat and humidity, excarnation can occur relatively quickly. Following death, bones lose water and the proteins begin to decompose to amino acids. Baylor researchers were able to follow these changes spectroscopically and correlate the data with the post-mortem interval by using regression modeling.
The Baylor researchers sampled 28 different pig bones that were up to 90-days old and used diffuse reflectance spectroscopy to try to date the skeletal remains. Busch said diffuse reflectance spectroscopy was used because it is non-destructive and is sensitive to protein and moisture. The bones were subjected to different light beams and the researchers found that the diffuse reflectance decreased as time moved on. The spectral data was then correlated with the known ages of the bones.
The researchers found the diffuse reflectance spectra of bones seemed to show some non-linearity in respect to changes with time. They decided to segment the data and divide it into three sets and make models for each. They found that by using the segmented approach, it would cut the prediction error substantially compared with the original 90-day model.
However this created a problem – which model should be used for an unknown bone sample? They found classifying the bones based on a discriminant analysis model followed by a segmented regression model gave the best results.
The research paper outlining the technique and results was presented at the annual meeting of the Federation of Analytical Chemistry and Spectroscopy Societies.