Author/Editor     Schweiger, Edvard
Title     Uporaba umetne inteligence pri vrednotenju vpliva klinično biokemičnih dejavnikov na farmakokinetiko učinkovin velike intersubjektne variabilnosti
Translated title     Application of artificial neural networks on evaluation of clinical biochemistry results on drugs' pharmacokinetic constants great intersubjective variability
Type     monografija
Place     Ljubljana
Publisher     Univerza v Ljubljani, Fakulteta za farmacijo
Publication year     2002
Volume     str. 99
Language     slo
Abstract     In the last decade usage of artificial network has increased. especially "feedforward/backpropagation neural networks". They are used all over the world in different areas, so it has been all a matter of time when it would be used in different pharmaceutical areas, like in bioequivalence studies to predict pharmacokinetic parameters. The constant problem of successful biocquivalence studies are drugs with known great intersubject and intrasubject variability. One of these drugs is nitrendipine product. In our study ww used thw artificial neural network because of its high practicability , feasibility and possibility to predict pharmacokinetic parameters for nitrondipine product to determinate the possible variability source and influence of clinical biochemistry results on pharmacokinetic parameters of nitrendipine. In our study of pharmacokinetic parameters we used AUC - area under curve, CMAX - peak plasma concentration, TMAX - time to reach plasma concentration and Vss/F volume of distribution, whereas for clinical biochemistry results we used height, body mass, creatinin serum conccntration, total bilirubin serum concentration, alanine aminotransferase (ALT) Ievel. gamma glutamile transferase (gamma-GT) level and albumin serum concentration. In creation of our database we included results gathered from a 2 x 4 double crossover randomize bioequivalence study of oral application nitrendipine products, involving in total 40 subjects and 160 applications. For the validation and for the test sample of neural network we excluded 15 applications or approximately 10% of all data set. For our calculations on our Windows 98 PC environment. we used the software product NNMODEL, and NNCALC for Microsoft Excel Add-In, by manufacturer Neural Fusion from New York. Modelling process involved the feedforward/backpropagation neural networks. sigmoid transfer function. (Abstract truncated at 2000 characters).
Descriptors     NEURAL NETWORKS (COMPUTER)
THERAPEUTIC EQUIVALENCY
NITRENDIPINE
CREATININE
BILIRUBIN
ALANINE AMINOTRANSFERASE
GAMMA-GLUTAMYLTRANSFERASE
SERUM ALBUMIN
BODY WEIGHT
BODY HEIGHT
ANALYSIS OF VARIANCE
RANDOMIZED CONTROLLED TRIALS