A statistical method that processes spectroscopic measurements very quickly could allow crime scene investigators to determine time of death of skeletal remains more accurately and quicker than before, according to researchers at Baylor University in Waco, Texas.

Once the flesh has rotted away, the skeleton remains, but there are few techniques that forensic scientists can call on to determine the so-called post-mortem interval precisely and quickly. Moreover, in places where it is hot and humid, the window of opportunity is smaller than ever as skeletalization occurs more rapidly.

The Baylor team, led by chemist Kenneth Busch, co-director of the Center for Analytical Spectroscopy, explains that bones lose water and the proteins within them decompose into their constituent amino acids. The team has tracked these changes using spectroscopy and then applied a statistical model, to correlate the spectra with the post-mortem interval. Their laboratory tests have an error rate as low as four days for bones that are 90 days old.

"In perfect conditions in the laboratory, the method looks very encouraging," explains Busch. "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."

The researchers used 28 different pig bones that were up to 90 days old, and used diffuse reflectance spectroscopy to obtain a date for the time of death of the skeletal remains. This spectroscopic technique is very sensitive to protein and moisture content. Moreover, it is a non-destructive technique, so that samples from the skeletal remains need not be removed.

The researchers found that the diffuse reflectance spectra of bones did not follow a straight line pattern as the bones age, so they segmented the data into three sets, which were then used to construct three statistical models of the aging process. They found that this approach could reduce the prediction error still further compared with the original 90-day model. A combination of the two approaches—a discriminant analysis model followed by a segmented regression model—gave the optimal results.

Busch and colleagues presented their technique and results to the annual meeting of the Federation of Analytical Chemistry and Spectroscopy Societies in October.

FACSS—Applications of Microscopy and Microanalysis in Forensic Science
Baylor University, Dr. K. Busch