This article discusses a research study that leveraged infrared spectroscopy along with machine learning models to differentiate between male and female Cochliomyia macellaria larvae, which are commonly found on human remains, offering a non-destructive approach to refine the supplemental post-mortem interval estimations and enhance accuracy of forensic analyses.
Forensic entomology plays an important role in medicolegal investigations by using insects, primarily flies, to estimate the time of colonization. This estimation relies on the development of the flies found at the (death) scene and can be affected (and sometimes corrected) by external factors, such as temperature and humidity, and internal factors, such as species and sex. This study leverages infrared (IR) spectroscopy combined with machine learning models—Partial Least Squares Discriminant Analysis (PLS-DA) and eXtreme Gradient Boosting trees Discriminant Analysis (XGBDA)—to differentiate between male and female Cochliomyia macellaria larvae, commonly found on human remains. Significant vibrational differences were detected in the infrared spectra of third instar C. macellaria larvae, with distinct peaks showing variations in relative absorbance between sexes, suggesting differences in biochemical compositions such as cuticular proteins and lipids. The application of PLS-DA and XGBDA yielded high classification accuracies of about 94 percent and 96 percent, respectively, with female spectra consistently having higher sensitivity than males. This non-destructive approach offers the potential to refine supplemental post-mortem interval estimations significantly, enhancing the accuracy of forensic analyses. (Published Abstract Provided)
Downloads
Similar Publications
- Development and Validation of a Method for Analysis of 25 Cannabinoids in Oral Fluid and Exhaled Breath Condensate
- Superhydrophobic Surface Modification of Polymer Microneedles Enables Fabrication of Multimodal Surface-Enhanced Raman Spectroscopy and Mass Spectrometry Substrates for Synthetic Drug Detection in Blood Plasma
- OpenSense: An Open-World Sensing Framework for Incremental Learning and Dynamic Sensor Scheduling on Embedded Edge Devices