Burnier Céline, Coulson Sally, Massonnet Geneviève, Pitts Kari, Sauzier Georgina, Lewis Simon W
Ecole des Sciences Criminelles, University of Lausanne, Switzerland.
Institute of Environmental Science and Research Ltd, 120 Mt Albert Road, Sandringham, Auckland 1025, New Zealand.
Sci Justice. 2021 May;61(3):235-248. doi: 10.1016/j.scijus.2021.01.005. Epub 2021 Feb 10.
Condom residues may be encountered in forensic investigations as traces in sexual assault or rape cases. Casework studies have shown the value of distinguishing condom residues from other types of personal products used by women. However, up to now, there has been no investigation of their chemical variability within an international context. This work employed attenuated total reflectance Fourier transform infrared spectroscopy with chemometrics to provide objective characterisation of condom lubricants and personal hygiene products from the international market. 166 samples were obtained covering five major classes of products likely to be used by women. Principal component analysis distinguished most major classes based on their spectral profiles, with subsequent support vector machine models yielding discrimination accuracies over 90%. A two-step approach was subsequently developed and enabled both classification and a discrimination accuracy of 100%. This could provide greater confidence in chemical discrimination of residues from these products when conducting investigations and help assess the origin of the chemical profile obtained. Further testing using three validation sets produced an identification accuracy of 100% for generic classes, which may allow investigative leads to be more readily obtained from recovered evidence.
在法医调查中，避孕套残留物可能会作为性侵犯或强奸案件中的痕迹被发现。实际案例研究表明，区分避孕套残留物与女性使用的其他类型个人用品具有重要意义。然而，到目前为止，尚未在国际范围内对其化学变异性进行研究。这项工作采用衰减全反射傅里叶变换红外光谱结合化学计量学方法，对国际市场上的避孕套润滑剂和个人卫生用品进行客观表征。共获取了166个样本，涵盖了女性可能使用的五大类产品。主成分分析根据光谱特征区分了大多数主要类别，随后的支持向量机模型判别准确率超过90%。随后开发了一种两步法，实现了100%的分类和判别准确率。这可以在调查过程中对这些产品的残留物进行化学鉴别时提供更大的信心，并有助于评估所获得化学特征的来源。使用三个验证集进行的进一步测试对通用类别产生了100%的识别准确率，这可能使调查线索更容易从回收的证据中获得。
