Pharmaceutical Analysis Research Aided By Chemometrics

Accomplished analysis of several pharmaceutical emphasized indifferent experiment phases used the Chemometrics as data process tool.The various Chemometrics methods in different phases included theOrthogonal Experimental Design method before pharmaceutical analysisexperiment; Nonlinear Least-squared Algorithms used for data-fitting ofexperiment result at the end of experiment; Back Propagation (BP)Artifical Neural Networks (ANN) used for a quantitativestructure-activity relationship (QSAR) model after experiment. The studyobjective were Pharmacokinetics of isosorbide-5-mononitrate (IS-5-MN),Reverse ion-pair chromatography retention mechanism of alkaloids,QSAR of Angiotensin Converting Enzyme Inhibitors (ACEI)respectively.The details are summarized as follows:1. Using orthogonal experimental design method got a optimizedElectrospray-Mass condition for isosorbide-5-mononitrate. Based it ananalytical method was developed for determination of IS-5-MN in humanplasma by high performance liquid chromatography/electrospray-mass(HPLC/ESI-MS) spectrometry. A NH₂ column was applied and Acyclovirwas used as internal standard in the procedure. The sample pretreated method was simple and results were accurate and precise. Method linearrange was 0.04-3.33μg/ml and RSD was 0.9982 respectively. Theanalytical method was successfully applied to the pharmacokinetics study.2. Using Nonlinear Least-squared Algorithms analyzedchromatography data of ten alkaloids in statistics. It shows that whenapplying perfluorinated carboxylic acid as ion-pair reagent, the retensionmechanism of ten alkaloids depend on the compositive associationconstant of the analyte ion, counter ion of ion-pair reagent and theproperty of the stationary ligand, which constant we define into N. Theperfluorinated carboxylic acid is suit to the LC-ESI-MS in a relativeconcentration for the analysis of alkaloids of nature product.3. The geometries and electronic structures of 20 ACEI had beenoptimized by using quantum chemical methods and 10 quantum-chemicalparameters such as energies, atom-charges et. al., were calculated. NineBP artificial neural networks with different nodes were trained to researchQSAR of ACEI, A 3-layers BP artificial neural network model with 6nodes in hidden layer was developed. The results of statistic analysis are:r²=0.995, S=0.050, which shows that artificial neural network model M6is more accurate and precise than multiple linear regression models. Theestablished ANN model can be applied to predict the activity of ACEItrustfully.