A Predictive Model for Toxicity Effects Assessment of Biotransformed Hepatic Drugs Using Iterative Sampling Method

Area of Science:

• Computational toxicology
• Drug development
• Machine learning applications

Background:

• Accurate toxicity prediction is crucial for efficient drug development.
• Existing computational models face challenges with imbalanced datasets and feature selection.
• A need exists for robust models to predict diverse toxicological endpoints.

Purpose of the Study:

• To develop and evaluate a novel computational model for predicting multiple drug toxicity effects.
• To enhance classification performance and reduce computational time through effective feature selection.
• To address the issue of imbalanced class distribution in toxicity datasets using advanced sampling techniques.

Main Methods:

• Feature selection using rough set-based methods.
• Implementation of an Iterative Sampling (ITS) method for imbalanced data, involving distribution modification and data cleaning.
• Classification using a Bagging classifier for predicting toxic or non-toxic drug samples.

Main Results:

• The proposed model demonstrated strong performance in classifying unknown drug samples across all four toxicity effects (mutagenic, tumorigenic, irritant, reproductive).
• Feature selection significantly improved classification efficiency and performance.
• The ITS method effectively handled imbalanced datasets, outperforming traditional sampling techniques.

Conclusions:

• The developed computational model offers a promising approach for predicting drug toxicity.
• The combination of rough set-based feature selection, ITS, and Bagging classifier provides an effective solution for imbalanced toxicity data.
• This model can aid in accelerating the drug development process by enabling early identification of potential toxic compounds.