Biospectroscopy with chemometrics analysis of blood for species identification and age estimation

• Main Author: Sandran, Durga Devi
• Format: Thesis
• Language: English
• Published: 2023
• Subjects: R Medicine; RA440-440.87 Study and teaching. Research; RC31-1245 Internal medicine
• Online Access: http://eprints.usm.my/61517/1/DURGA%20DEVI%20AP%20SANDRAN-TESIS%20P-SKD001517-E.pdf; http://eprints.usm.my/61517/
• Institution: Universiti Sains Malaysia

Accurate species identification and age estimation of bloodstains are indispensable tools in forensic investigations, providing invaluable information that can significantly impact case outcomes. However, existing methods are challenging due to their destructive and strenuous procedures. This thesis explores the effectiveness of biospectroscopic techniques and multivariate chemometrics integration for species and age determination of bloodstains, using samples from eight species: humans, cattle, chicken, deer, duck, fish, goat, and swine. The present study deployed tandem analysis of attenuated total reflection-Fourier transform infrared (ATR-FTIR) and ultraviolet-visible (UV/Vis) spectroscopy, complemented by principal component analysis and linear discriminant analysis (PCA-LDA) to determine bloodstain species. The PCA-LDA models for infrared and visible spectroscopy were built and showed complete differentiation between human and animal spectra. The results demonstrated that the ATR-FTIR spectroscopy significantly outperformed visible spectroscopy in discriminating human and animal blood. With a classification accuracy of 98.3%, the infrared PCA-LDA model effectively distinguished between various species, including a perfect classification for humans, chickens, cattle, ducks, and fish. For in-situ bloodstain age analysis, human and animal blood spots were prepared and stored under two conditions: indoors and outdoors. The blood spots were deposited on ten porous and non-porous substrates over one year to simulate a crime scene. Utilising ATR-FTIR spectroscopy, two major blood protein bands (Amides I and II) can be successfully detected in fresh and aged blood spectra for up to one year. A total of 160 partial least squares regression (PLSR) models were developed, with superior predictive performance observed for outdoor bloodstains (RMSE: ~0.29-2.42; R2: ~0.56-0.99) compared to indoor ones (RMSE: ~0.51-3.28; R2: ~0.20-0.98). Lower predictive Root Mean Squared Error (RMSE) and high R² scores for bloodstains on all ten substrates, irrespective of species, enhanced this research's practicality. These models were further applied to create partial least squares-discriminant analysis (PLS-DA) models, demonstrating outstanding categorisation ability up to ~99% (indoors) and ~98% (outdoors) for aged blood spots on porous and non-porous surfaces. Both validation results showed prominent potential and statistically reliable discrimination for estimating the bloodstain age on various substrates, notably for outdoor and longer ageing terms. In summary, this research proves that integrating ATR-FTIR spectroscopy with multivariate chemometrics provides a non-invasive, conclusive and rapid strategy for determining the species and age of bloodstains in forensic investigations with promising practical applications in real cases benefiting the forensic and law enforcement communities.