Spectroscopy and chemometrics for differentiation of sex in human head hair fibers

Human hair is one of the most common pieces of evidence in forensics, however, it lacks a non-destructive, timeefficient, methodology for qualitative information. With the already well-established method for fiber analysis in spectroscopy and the common presence of Fourier Transform Infrared (FTIR) and microspectrophotometry (MSP) instrumentation in forensics, this study proposes chemometric approach to differentiate hair fibers on the sex assigned at birth. The IR and MSP hair sample spectra from different sexes were collected and, following spectral pre-treatment and mid-level data fusion, were used to calibrate a neural network model with 99.6% accuracy and 99.4% validation accuracy through an external validation dataset. This approach provides a robust model with significant qualitative information for forensic investigations that can be easily applied in forensic labs across the world. These results represent a rather small population size which would need to be expanded for real-world forensic applications.