Contribution de l’Intelligence Arfificielle et de la chimie Quantique à la Modélisation des propriétés ADMET des Dérivés Thiazolidinones

Abstract

This work aims to investigate thiazolidinone derivatives and evaluate their drug potential through the integration of computational chemistry and machine learning techniques. Density Functional Theory (DFT) calculations were performed to extract important quantum chemical descriptors, including HOMO and LUMO energies and the energy gap, in addition to physicochemical descriptors such as LogP, TPSA, and molecular weight. The pharmacokinetic and drug-likeness properties of these compounds were also evaluated using ADMET parameters and Lipinski’s Rule of Five. The study employed Quantitative Structure–Activity Relationship (QSAR) modeling combined with artificial intelligence and machine learning techniques to develop predictive models capable of correlating molecular descriptors with ADMET properties. The resu lts revealed clear relationships between molecular structure and the electronic and pharmacological properties of the investigated compounds. Furthermore, several molecular descriptors were identified as key factors influencing pharmacokinetic behavior and bioavailability. The findings demonstrate the effectiveness of combining DFT calculations with machine learning approaches for predicting the pharmacological properties of thiazolidinone derivatives, thereby accelerating the drug discovery process and reducing experimental costs. Moreover, the study enabled the identification of several promising compounds that may serve as potential candidates for the development of new drugs with high therapeutic efficacy and low toxicity.