Feature Selection and Rapid Characterization of Bloodstains on Different Substrates

Area of Science:

• Forensic Science
• Analytical Chemistry
• Biophysics

Background:

• Crime scene investigation relies on accurate timelines, but current methods for determining bloodstain age are limited and destructive.
• Raman spectroscopy offers a nondestructive approach for analyzing body fluids, providing molecular insights without sample processing.
• Existing Raman spectroscopy models struggle with substrate interference, limiting their application to pure body fluids for age estimation.

Purpose of the Study:

• To develop a Raman spectroscopy method for accurately estimating bloodstain age on diverse substrates.
• To identify selective spectral markers for bloodstain age determination despite substrate interference.
• To create a robust and scalable model for real-world forensic applications.

Main Methods:

• Utilized Raman spectroscopy to analyze bloodstains aged on various substrates (floor tile, facial tissue, linoleum-polymer).
• Employed Least Absolute Shrinkage and Selection Operator (LASSO) regression for automated feature selection to manage substrate interference.
• Trained substrate-specific models using pure bloodstain spectra to predict the age of composite spectra.

Main Results:

• Achieved low predictive root mean square error (0.204–0.222 h in logarithmic scale) and high R-squared values (0.913–0.926) on test data.
• Demonstrated the model's ability to accurately predict bloodstain age on different surfaces by overcoming substrate signal interference.
• Validated a framework for automated feature selection enabling accurate age estimation from composite bloodstain spectra.

Conclusions:

• Raman spectroscopy combined with LASSO regression provides a reliable, nondestructive method for determining bloodstain age on various surfaces.
• The developed automated feature selection algorithm allows for the creation of easily applicable and scalable substrate-specific models.
• This technique significantly aids forensic investigations by enabling precise timeline establishment at crime scenes.