Computational intelligence analysis on drug solubility using thermodynamics and interaction mechanism via models...

Area of Science:

• Pharmaceutical Science
• Computational Chemistry
• Data Science

Background:

• Accurate prediction of drug solubility and API-polymer interactions is crucial for drug development.
• Complex datasets present challenges for traditional predictive modeling techniques.

Purpose of the Study:

• To evaluate and compare the performance of various regression models for predicting drug solubility and API-polymer interactions.
• To assess the efficacy of the Fireworks Algorithm (FWA) for hyper-parameter tuning in these models.

Main Methods:

• Investigated Gaussian Process Regression (GPR), Support Vector Regression (SVR), Bayesian Ridge Regression (BRR), and Kernel Ridge Regression (KRR).
• Applied Z-score preprocessing for outlier detection and data refinement.
• Utilized the Fireworks Algorithm (FWA) for hyper-parameter optimization.

Main Results:

• Gaussian Process Regression (GPR) demonstrated superior performance, achieving the highest R-squared values (0.9980 for training, 0.9950 for testing) and lowest Mean Squared Error (MSE) and Mean Absolute Error (MAE).
• The Fireworks Algorithm (FWA) effectively enhanced the predictive capabilities of the regression models.
• Z-score preprocessing improved data accuracy and analysis reliability.

Conclusions:

• Gaussian Process Regression (GPR) is a robust and precise method for complex regression tasks in pharmaceutical and scientific fields.
• The Fireworks Algorithm (FWA) shows significant potential for optimizing predictive models.
• Selecting appropriate regression models and optimization techniques is vital for dependable predictive analytics.