ML-Powered MultiFluid-mRNAplex31: A Self-Developed Panel for Precise Tracing of Five Body Fluids

Identification of body fluid origin at crime scenes can help establish a link between the individual who sheds the sample and the criminal activity, thereby aiding the characterization of the case. Many studies have shown that messenger RNA (mRNA) expression profiling is a sensitive and reliable method for body fluid origin identification. However, it still faces challenges such as false positive results due to the cross-reactivity of specific markers. To address these issues, we developed MultiFluid-mRNAplex31, an mRNA multiplex amplification system designed to distinguish common single-source and mixed body fluids in forensic molecular biology. The system integrates 26 body fluid-specific mRNAs expressed differentially among blood, saliva, vaginal fluid, menstrual blood, and semen (at least four to six specific mRNAs for each body fluid) alongside two sex-specific markers and three housekeeping genes. The multiplex amplification detection system was constructed using reverse transcription-polymerase chain reaction (RT-PCR) combined with capillary electrophoresis (CE) technology, and its forensic performance was comprehensively evaluated. Evaluation results demonstrated exceptional species specificity, sensitivity, and robustness in analyzing aged and mixed samples. Based on this system, we developed four types of body fluid classification prediction models [random forest, support vector machine (SVM), neural network (NN), and CatBoost], which accurately identified five types of body fluids in single-source biological samples. To extend this capability to mixed samples, we prepared 17 different types of mixtures and developed an array of machine-learning models capable of handling single-source fluids and mixtures. After evaluating the performance of the classifiers over 1000 validation iterations, we found that the average accuracy of SVM and NN models exceeded 0.96. This study establishes a comprehensive analysis system for precise body fluid identification, providing strong technical capabilities for forensic science.