EXTRACTING THE CAUSE-EFFECT RELATIONSHIPS FROM FORMULATION DATA OF CONTROLLED RELEASE PRODUCT USING VISUALISATION METHOD

Abstract

Background: The study of examining the influence of ingredients to the properties of the product especially the release of active ingredients is regular job and is very important. This paper presents a method to support surveying the cause-effect relationships from experimental data of controlled release tablet formulation with an intuitive method - parallel coordinates. Materials and methods: In the survey of cause-effect relationships, the experimental data of tablets controlled release formulation referenced from a published article and parallel techniques will be applied to extracting the intuitive relationships between the variables of ingredients and properties. Results: With the parallel coordinates technique combined to neural network for modeling formulation data, the formulator can manipulate to change the value of independent variable X by selecting the values ​​to be examined to see the change of the dependent variable Y. Compared with the predicted quadratic form from the published result, parallel coordinates view has taken the cause and effect relationship and a more accurate visualisation. Conclusion: This result will lead formulators to understanding their products more precisely and saving a lot of time and labor in R&D process. Keywords: neural networks, visualisation, cause-effect relationships, controlled release Key words: neural networks, visualisation, cause-effect relationships, controlled release