Abstract：
In forensics, it is important to determine the time since deposition (TSD) of bloodstains, one of the most common types of biological evidence in criminal cases. However, no effective TSD inference methods have been established despite extensive attempts in forensic science. Our study investigated the changes in the blood transcriptome over time, and we found that degradation could be divided into four stages (days 0-2, 4-14, 21-56, and 84-168) at 4 °C. A random forest prediction model based on these transcriptional changes was trained on experimental samples and tested in separate test samples. This model was able to successfully predict TSD (area under the curve [AUC] = 0.995, precision = 1, and recall = 1). Thus, this proof-of-concept pilot study has practical significance for assessing physical evidence. Meanwhile, 11 upregulated and 13 downregulated transcripts were identified as potential time-marker transcripts, laying a foundation for further development of TSD analysis methods in forensic science and crime scene investigation.
摘要：
在取证中，确定血迹沉积（TSD）以来的时间很重要，刑事案件中最常见的生物证据之一。然而,尽管在法医学中进行了广泛的尝试，但尚未建立有效的TSD推断方法。我们的研究调查了血液转录组随时间的变化，我们发现在4°C下降解可分为四个阶段（第0-2、4-14、21-56和84-168天）。在实验样本上训练基于这些转录变化的随机森林预测模型,并在单独的测试样本中进行测试。该模型能够成功预测TSD（曲线下面积[AUC]=0.995，精度=1，召回率=1）。因此，这项概念验证试点研究对评估实物证据具有现实意义。同时,11个上调和13个下调的转录本被鉴定为潜在的时间标记转录本，为进一步发展法医学和犯罪现场调查中的TSD分析方法奠定了基础。
