Prediction of the partition coefficients using QSPR modeling and simulation of paclitaxel release from the diffusion-controlled drug delivery devices.

Abstract

An in silico approach is proposed to first predict the partition coefficient of the model drug, paclitaxel, in different biocompatible and biodegradable polymer versus the blood plasma using artificial neural networks (ANNs) and semi-empirical quantitative structure property relationships (QSPRs). A simplified molecular-input line-entry system (SMILES) notation is used to represent the structures of the different polymers and the drug. The SMILES notation is then used to calculate the various structure-based descriptors. These descriptors are then used in the ANNs and semi-empirical QSPRs to predict the properties for a given drug-polymer device. A fluid flow model is subsequently solved to simulate the controlled drug release in the blood plasma. The effects of various parameters are also studied on the drug release profiles from these devices. The proposed approach provides a systematic framework to simulate the controlled release of the drug from the diffusion-controlled drug-polymer release systems. The developed models can be used in a reverse engineer framework to design the controlled delivery devices for a target drug release profile in near future.