Class identity assignment for amphetamines using neural networks and GC-FTIR data.

An exploratory analysis was performed in order to evaluate the feasibility of building of neural network (NN) systems automating the identification of amphetamines necessary in the investigation of drugs of abuse for epidemiological, clinical and forensic purposes. A first neural network system was built to distinguish between amphetamines and nonamphetamines. A second, more refined system, aimed to the recognition of amphetamines according to their toxicological activity (stimulant amphetamines, hallucinogenic amphetamines, nonamphetamines). Both systems proved that discrimination between amphetamines and nonamphetamines, as well as between stimulants, hallucinogens and nonamphetamines is possible (83.44% and 85.71% correct classification rate, respectively). The spectroscopic interpretation of the 40 most important input variables (GC-FTIR absorption intensities) shows that the modeling power of an input variable seems to be correlated with the stability and not with the intensity of the spectral interaction. Thus, discarding variables only because they correspond to spectral windows with weak absorptions does not seem be not advisable.

Structure-activity correlations for illicit amphetamines using ANN and constitutional descriptors.

The goal of this study was to develop an expert system capable to identify the potential biological activity of new substances having a molecular structure similar to illicit amphetamines. For this purpose we have designed two types of artificial neural network (ANN) systems, which have been trained to classify amphetamines according to their toxicological activity (stimulant amphetamines or hallucinogenic amphetamines) and distinguish them from nonamphetamines. Such a system is essential for testing new molecular structures for epidemiological, clinical, and forensic purposes. The first type of artificial neural network is a "spectral" neural network, which has as input variables the most important 100 absorption intensities from a total of 260 measured for each normalized infrared spectrum 10cm(-1) apart. The spectral data consists of a database built with the GC-FT-IR spectra of the most popular drugs of abuse (mainly central stimulants, hallucinogens, sympathomimetic amines, narcotics and other potent analgesics), precursors and derivatized counterparts. All samples were also characterized by their constitutional descriptors (CDs). For each sample, a number of 45 CDs were computed and introduced as input variables for a second type of ANN, which uses a structural database. The efficiency of this "structural" artificial neural network (CD-ANN) has been improved by optimizing the training set and increasing the number of input variables (CDs). A comparative analysis of the spectral and the structural networks is presented.