Estimating post-mortem interval (PMI) through purine analysis in muscle tissue using extreme gradient boosting

Post-mortem biochemical changes in muscle tissue can serve as valuable indicators for estimating the post-mortem interval (PMI), following time and temperature-dependent patterns. This study focused on purine compounds: adenosine, inosine, hypoxanthine, and xanthine as potential biochemical markers of PMI.Muscle tissue samples were collected from Sus scrofa domesticus (domestic pigs) and were stored at 10 °C, 20 °C, and 30 °C and then analysed over 8 consecutive days. Concentrations of four purines were measured throughout this period to examine their post-mortem kinetics. Using these data, we developed predictive machine learning model to assess PMI based on Extreme Gradient Boosting (XGBoost), a machine learning algorithm capable of modelling complex temporal trends.Among the compounds, inosine, hypoxanthine, and xanthine showed clear and consistent degradation patterns, while adenosine remained relatively stable. Our results show that by using multiple consecutive post-mortem measurements, it significantly improved the model's predictive accuracy, highlighting the value of short-term repeated sampling.These findings support the potential of combining purine analysis with machine learning to estimate PMI and demonstrate that short sequential sampling of those compounds can be useful for forensic applications even though validation under more variable post-mortem conditions is still needed.