A new approach to assess post-mortem interval: A machine learning-assisted label-free ATR-FTIR analysis of human vitreous humor

Abstract

A crucial issue in forensics is determining the post-mortem interval (PMI), the time between death and the finding of a body. Despite various methods already employed for its estimation, only approximate values are currently achievable. Vitreous humor (VH) is an avascular tissue between the lens and the retina, mainly composed by a collagen fibers network, hyaluronic acid, and hyalocytes. Recently, it has received interest in forensic medicine, being easy to collect and susceptible to low microbiological contamination and putrefaction. Based on this evidence and thanks to the ability of Attenuated Total Reflectance – Fourier Transform InfraRed (ATR–FTIR) spectroscopy to perform fast analyses on a minimal sample amount, in this study, a new analytical approach to reliably estimate PMI is proposed combining ATR-FTIR analysis of VH human samples with multivariate statistical procedures, such as Principal Component Analysis (PCA) and Partial Least Squares-Discriminant Analysis (PLS-DA), for discriminant classification. Regression procedures, including Partial Least Squares Regression (PLSR), were performed: extremely positive results were obtained, and the most discriminant spectral features were highlighted (peaks at 1665, 1630, 1585, 1400, 1220, 1200, 1120, 854, 835, and 740 cm⁻¹) and associated to PMI classes (average accuracy over 80 %). Specific and reliable markers able to correlate the macromolecular composition of VH with the PMI were identified, revealing a post-mortem protein degradation and amino acids deamination (decrease of proteins and increase of free amino acids and NH3), an increase of lactate, which diffuses from the retina to the VH, and changes in the hyaluronic acid component.