Chemical profiling and classification of illicit heroin by principal component analysis, calculation of inter sample correlation and artificial neural networks.

摘要: Artificial neural networks (ANNs) were utilised to validate illicit drug classification in the profiling method used at "Institut de Police Scientifique" of University Lausanne (IPS). This established links between samples using a combination principal component analysis (PCA) and calculation correlation value samples. Heroin seizures sent IPS laboratory analysed gas chromatography (GC) separate major alkaloids present heroin. Statistical was then performed on 3371 Initially, PCA as preliminary screen identify similar chemical profile. A calculated for each sample previously identified with PCA. determine These recorded an Ibase((R)) database. From this database notion "chemical class" arises, where profiles are grouped together. Currently, about 20 classes" have been identified. The normalised peak areas six target compounds train ANN classify into its appropriate class. Four hundred sixty-eight training data set. Sixty treated blinds 370 non-linked results show that 96% cases network attributed seizure right class". application found be useful tool new existing classes. should increasingly such situations involving profile comparisons classifications.