Dataset for: Portable Fourier-transform infrared spectroscopy and machine learning for sex determination in third instar Chrysomya rufifacies larvae

Description

Abstract

Forensic entomology is crucial in medicolegal investigations, utilizing insects—primarily flies—to estimate a supplemental post-mortem interval based on their development at the (death) scene. This estimation can be influenced by extrinsic factors like temperature and humidity, as well as intrinsic factors such as species and sex. Previously, benchtop Fourier-transform infrared (FTIR) spectroscopy coupled with machine learning demonstrated high accuracy in distinguishing the sex of third instar Cochliomyia macellaria larvae. This study leverages benchtop- and handheld-based FTIR spectroscopy combined with machine learning models—Partial Least Squares Discriminant Analysis (PLSDA), eXtreme Gradient Boosting trees Discriminant Analysis (XGBDA), and Artificial Neural Networks Discriminant Analysis (ANNDA)—to differentiate between male and female Chrysomya rufifacies larvae, commonly found on human remains. Significant vibrational differences were detected in the mid-infrared spectra of third instar Ch. rufifacies larvae, with a majority of peaks showing a higher abundance of proteins, lipids, and hydrocarbons in male larvae. PLSDA and ANNDA models developed using benchtop FTIR data achieved high external validation accuracies of approximately 90% and 94.5%, respectively, when tested with handheld FTIR data. This nondestructive approach offers the potential to refine supplemental post-mortem interval estimations significantly, enhancing the accuracy of forensic analyses of entomological evidence.