Privatsphäre-Einstellungen

Vorhersage der Eigenschaften verschiedener Tablettenformate unter Verwendung künstlicher neuronaler Netze

Jan-Niklas Francke, Robert Frank Lammens und Klaus-Jürgen Steffens

Pharmazeutisches Institut, Pharmazeutische Technologie, Rheinische Friedrich-Wilhelms-Universität, Bonn

Korrespondenz: Prof. Dr. Klaus-Jürgen Steffens, Pharmazeutisches Institut, Pharmazeutische Technologie, Rheinische Friedrich-Wilhelms-Universität Bonn, Gerhard-Domagk-Str. 3, 53121 Bonn (Germany), e-mail: steffens@uni-bonn.de

Prediction of the properties of various shaped tablets using artificial neural networks (ANN)

The production of various shaped tablets is advantageous for pharmaceutical manufacturers concerning subsequent processing, optimisation of product characteristics and identification of the dosage form. Typically, small sized tablets are analysed for the purpose of development using minimum amounts of formulation, whereas production shapes differ considerably in dimensions and profiles.
Among this challenging task, a model is presented which facilitates the prediction of porosity and mechanical strength of convex-faced or capsule-shaped tablets based on a batch of small cylindrical compacts. To elucidate the influence of geometrical variability on tablet properties, artificial neural networks have been used for prediction. This promising methodology is considered to be able to map multidimensional, nonlinear relationships between process parameters and quality characteristics of tablets.

Key words Bruchfestigkeit • Direkttablettierung • Künstliche Neuronale Netze (KNN) • Porosität • Tablettenformat