Investigation of semen deposition time: Spectral analysis using Fourier transform infrared spectroscopy at different temperatures in vivo

For forensic application, except for the detection of DNA, there remains a significant gap in the field of time estimation for semen traces. We utilized Fourier Transform Infrared (FTIR) technology and chemometric methods to investigate the time since deposition of semen in vivo groups at different temperatures. Principal component analysis revealed that the degradation patterns of semen were divided into three stages, primarily dominated by moisture, proteins, and degraded small molecules, respectively. Additionally, within the time frame of 0–168 h, robust Partial Least Squares Regression (PLSR) and Random Forest (RF) time prediction models with R² > 0.9 were successfully established in all three environmental conditions, with the best Root Mean Square Error of Prediction values being 7.1796 and 4.1359, respectively. The key variables for both the PLSR and RF models were primarily concentrated in the wavenumber ranges of 1000–1180, 1320–1480, 1530–1630, and 1700–1760 cm⁻¹ . Furthermore, the PLSR model, after multivariate scatter correction preprocessing, was suitable for scenarios involving prolonged time and high temperatures that lead to more thorough degradation of semen, while the RF model demonstrated valuable predictive performance under opposite conditions. The results indicate that the use of FTIR and chemometric methods can further elucidate temporal and procedural details in sexual offenses, showcasing significant potential in the investigation of actual cases.