Non-invasive prediction of bloodstain age using the principal component and a back propagation artificial neural network

Abstract

The age determination of bloodstains is an important and immediate challenge for forensic science. No reliable methods are currently available for estimating the age of bloodstains. Here we report a method for determining the age of bloodstains at different storage temperatures. Bloodstains were stored at 37 °C, 25 °C, 4 °C, and −20 °C for 80 d. Bloodstains were measured using Raman spectroscopy at various time points. The principal component and a back propagation artificial neural network model were then established for estimating the age of the bloodstains. The results were ideal; the square of correlation coefficient was up to 0.99 (R 2 > 0.99) and the root mean square error of the prediction at lowest reached 55.9829 h. This method is real-time, non-invasive, non-destructive and highly efficiency. It may well prove that Raman spectroscopy is a promising tool for the estimation of the age of bloodstains.

An Exploratory Time Since Deposition Analysis of Whole Blood Using Metrics of DNA Degradation and Visible Absorbance Spectroscopy

“…Established techniques for age estimation of forensic evidence uses a combination of approaches housed in entomology, anthropology and taphonomy (Amendt et al 2007;Iqbal et al 2017;Wescott 2018). Research in these areas provide valuable insight into meteorological variables that have significant influence on the accuracy of such aging techniques, such as temperature and humidity (Bremmer et al 2011;Sun et al 2017). There is; however, a growing body of scientific literature that is developing predictive models for age based on the spectrometric characteristics of macromolecule degradation.…”

The crux of time: A meta-analysis of ex vivo whole blood degradation

“…Commonly analyzed biomolecules for TSD estimation models include hemoglobin (Hb) from red blood cells (RBCs) and genetic material from white blood cells (WBCs). Spectroscopic techniques, such as Raman (Premasiri et al, 2012;Doty et al, 2016Doty et al, , 2017Sun et al, 2017b;Takamura et al, 2019;Menżyk et al, 2020), infrared (Botoniic-Sehic et al, 2009;Lin et al, 2017;Zhang et al, 2017;Hassan et al, 2019), reflectance (Bremmer et al, 2011a(Bremmer et al, , 2011bLi et al, 2011;Edelman et al, 2012aEdelman et al, , 2016Sun et al, 2017a) and UV-Vis absorbance (Hanson and Ballantyne, 2010;Hanson et al, 2011;Agudelo et al, 2015;Bergmann et al, 2017Bergmann et al, , 2021Kaur et al, 2020;Stotesbury et al, 2020;Cossette et al, 2021) spectroscopies have all probed the timewise degradation of Hb. Photography (Thanakiatkrai et al, 2013;Shin et al, 2017;Choi et al, 2019) and hyperspectral imaging (Edelman et al, 2012b;Li et al, 2013;Majda et al, 2018) have measured the colour changes in blood while electron spin resonance and electrochemistry analyzed the conformational changes of the heme groups (Miki et al, 1987;Matsuoka et al, 1995;Fujita et al, 2005).…”

The crux of time: A meta-analysis of ex vivo whole blood degradation

“…A variety of environmental conditions have been studied in bloodstain TSD research; this includes exposure to outdoor environments 12,30,71,72,82 , varying temperatures (-20 to 61 °C) and a wide range of humidity values (10 to 99% RH) 9,39,[71][72][73]79 . Statistical analyses also differ between studies; regressions are common, but the type of fit varies (linear 78 , polynomial 51 , logarithmic 28 , exponential 88 , multivariate 39 ) depending upon the observed time-series trends. A number of studies used dimensionality-reduction tools for their analyses 24,36 while others incorporated cross-validation to obtain prediction accuracies 33,45 .…”