| Hydroxypropylmethylcelluose(HPMC) hydrophilic matrix systems areamong the most widely used means for controlled drug delivery in solid oral dosage. It is because that HPMC matrix tablets not only has the merits of controlled release systems but also has the advantage of ease manufacture ,good sustained release effect ,their physical character unconcerned with pH in physiological range etc. it was well reported about the mechanism of drug release from HPMC matrix tablets and the variables that influence the drug release from matrix tablets. The factors influence the release of drugs from HPMC hydrophilic matrices include internal and external factors, the internal involve the amount of HPMC ,viscosity of polymer ,mixture of polymer ,the property of drug and filling ,the drug dose, the amount of filling the thickness of tablets ,particle etc, the external factors were the process for produce the tablets, the property of media ,such as ionic strength ,pH ,the stirring speed of dissolution machineThe factors that influence drugs release from HPMC matrix tablets are composed of formulation and process factors, more important, there is no line correction among these factors. The classic way to deal with the no-line correction question is RSM(Response Surface Methodology).however, since prediction of pharmaceutical responses based on the quadratic polynomial is often limited to low levels. It may result in the poor estimation of optimal formulations. To overcome the shortcomings of the poor estimation of RSM based on the quadratic polynomial, application of an artificial neural network (ANN) has been investigated [24-30].Artificial neural networks are computational paradigms that can simulate the neurological processing ability of the human brain. The neural networks, consisting of inter- connected adaptive processing units, so-called "neurons", are able to discern complex and latent patterns in the information presented to them. Their adaptive nature makes such computational models very appealing in application domains where one has a poor or incomplete understanding of the problem to be solved but where training data is readily available. Neural networks are applicable to solving a wide variety of problems, including pattern classification, function approximation, associative memory, clustering, forecasting and prediction, combinatorial optimization, nonlinear system modeling, and control.In my theme ,at first we decided to use ANN to aid to design two-factor HPMC matrix tablets: we chose poor soluble drug allopurinol as model drug ,fix the other variables, the amount of HPMC in each tablet and intrinsic viscosity as inputs variables ,the accumulated drug release in each different sampling time were used as output responses, the 14 formulations in 17 preparation formulations were used as training set, while the other left 3 formulations were used as validating formulations, "leave one out" cross validation were used as training approach for the ANN, and ,compared with RSM(response surface methodology), predict the amount of HPMC and intrinsic viscosity of HPMC on drug release in 2,4,6,8 hours, then we used the 3-dimention diagram to demonstrate the influence of the amount of HPMC and the intrinsic viscosity of HPMC on drug release; the regression equation and siminarity factors show the reliability of ANNs agree with RSM.,the influence of the amout of HPMC and its intrinsic viscosity on drug release agree with the literatures.we chose the insoluble drug Flurbiprofen as model drug ,the pH 7.4 PBS as media, the amount of HPMC in each tablet and the stirring speed of dissolution machine as casual factors ,named inputs ,the accumulated drug release in each different sampling time were used as output responds ,the 13 formulations in 18preparation formulations were used as training set ,the other 5 formulations were used as validated formulations , "leave one out" cross validation were used as training approach for the ANN , we compared the ANN with RSM, depending on the trained ANN and RSM, we predict...
|
PDF Full Text Request