Computational estimation of drug’s concentration distribution through a microporous membrane using artificial intelligence approach

Abstract

Recently, the emergence of promising membrane-based processes in disparate related disciplines of pharmaceutical industries (i.e., biochemical engineering and biotechnology) has paved the way towards concentration and purification of therapeutic entities, molecular synthesis, and drug delivery. The strategy of this investigation is to employ various artificial intelligence (AI)-based computational models as a novel method to estimate the concentration distribution value of a medical solute inside the membrane. To achieve this, a data set with two inputs and one output was analyzed using bagging ensemble method with some distinct base models. The base models are Multi-layer Perceptron (MLP), Linear Regression (LR), and K-Nearest Neighbor (KNN). The hyperparameters of the methods were developed through performing a grid search and the final models were examined. Based on this, as expected, the use of bagging along with the optimization of hyperparameters led to strong models. All three bagging models with the R-square measure have scores higher than 0.98. Also, in terms of RMSE bagging LR, bagging MLP, and bagging KNN have error rates of 9.25 × 101, 2.29 × 101, 9.57 × 101, respectively.