Abstract/Summary

A suitable regression model for predicting the dissolution profile of Poly (lactic-co-glycolic acid) (PLGA) micro-and nanoparticles can play a significant role in pharmaceutical/medical applications. The rate of dissolution of proteins is influenced by several factors and taking all such influencing factors into account; we have a dataset in hand with three hundred input features. Therefore, a primary approach before identifying a regression model is to reduce the dimensionality of the dataset at hand. On the one hand, we have adopted Backward Elimination Feature selection techniques for an exhaustive analysis of the predictability of each combination of features. On the other hand, several linear and non-linear feature extraction methods are used in order to extract a new set of features out of the available dataset. A comprehensive experimental analysis for the selection or extraction of features and identification of the corresponding prediction model is offered. The designed experiment and prediction models offer substantially better performance over the earlier proposed prediction models in literature for the said problem.