Theoretical analysis of MOFs for pharmaceutical applications by using machine learning models to predict loading...

Area of Science:

• Materials Science
• Nanotechnology
• Computational Chemistry

Background:

• Metal-organic frameworks (MOFs) possess porous structures ideal for drug delivery applications.
• Accurate prediction of drug loading capacity and cytotoxicity is crucial for MOF development.
• Machine learning offers a powerful tool for analyzing complex MOF properties.

Purpose of the Study:

• To develop and evaluate a machine learning model for predicting drug loading capacity and cell viability in MOFs.
• To assess the efficacy of a stacking regression approach for optimizing MOF-based drug delivery systems.
• To provide insights into the chemical and biological applications of MOFs through predictive modeling.

Main Methods:

• A stacking regression framework combining Multilayer Perceptron (MLP), Random Forest (RF), and Quantile Regression (QR) was employed.
• Principal Component Analysis (PCA) was utilized for dimensionality reduction.
• The Water Cycle Algorithm was used for hyperparameter optimization.

Main Results:

• The Quantile Regression-MLP (QR-MLP) stacking model demonstrated superior performance.
• Achieved high R-squared (R²) scores: 0.99917 for Drug Loading Capacity and 0.99111 for Cell Viability.
• The model effectively handled complex datasets, showcasing the robustness of stacking ensemble methods.

Conclusions:

• Stacking ensemble approaches are highly efficient for analyzing MOF properties and optimizing drug delivery systems.
• The developed QR-MLP model provides a reliable method for predicting critical MOF performance metrics.
• Findings suggest significant potential for advancing MOF applications in drug delivery and other biological uses.