Coupling the image analysis and the artificial neural networks to predict a mixing time of a pharmaceutical powder

Abstract

In recent years, different laboratories were interested in predicting the mixing time of a pharmaceutical powder. In fact, a nonhomogeneous mixture may lead to under dose and/or overdose of the active ingredient in the drug product. Our study is aimed toward using a new and revolutionary approach in the field of the processes "The Artificial Neural Networks" (ANN) by using the Neural Networks Toolbox TM derived from Matlab ® software. The validation of the neural network was assumed by studying others mixing p o w d e r s and then w e compared the experimental results to the data obtained by the neural network calculations. Experimental results were obtained from a non-destructive method (Image Analysis) which was used in order to characterize the homogeneity of powder mixture in a V-Blender as well as a Cubic Blender which are most used in the pharmaceutical industry. Keywords

Applications of Artificial Intelligence and Machine Learning Algorithms to Crystallization

“…For instance, Kumar et al 177 used four ML algorithms trained with discrete element simulations to develop a protocol to ensure uniform solid bed mixing of pharmaceutical powders in agitated filter dryers, finding regression tree random forest giving the best accuracy. Mahdi et al 272 predicted the mixing time of powders in cubic V-blender mixers with relative errors lower than 4% using an ANN. Finally, for a recent comprehensible review on machine learning for drug discovery focusing on bioactivity, we refer to Carracedo-Reboredo et al, 178 where 0D to 4D molecular descriptors and available pharmaceutical databases are discussed.…”

Image Analysis: A Versatile Tool in the Manufacturing and Quality Control of Pharmaceutical Dosage Forms

“…To predict the required time for mixing pharmaceutical powders, Mahdi et al used a new approach. They coupled the artificial neural networks with image analysis for the characterization of the homogeneity of powder mixtures [ 34 ]. Berthiaux et al developed an IA-based method for the measurement of powder homogeneity of loose materials and proposed their real-time principal component analysis as an on-line methodology adaptable to other techniques [ 35 ].…”

Optimization of the artificial neuronal network for the degradation and mineralization of amoxicillin photoinduced by the complex ferrioxalate with a gradual and progressive approach of the ligand

“…The performance of the Feed Forward Back Propagation and Cascade Forward Back Propagation was evaluated using the mean squared error (MSE) and / or correlation coefficient (r) [42][43][44].…”