Machine learning analysis of oral solid dosage formulation solubility variations by adjusting pressure and...

Area of Science:

• Computational chemistry and materials science.
• Application of machine learning in chemical engineering.

Background:

• Accurate prediction of chemical properties like solubility and density is crucial for industrial processes.
• Supercritical carbon dioxide (SC-CO2) is a key solvent in various chemical and pharmaceutical applications.
• Tolfenamic acid solubility is important for drug formulation and delivery.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting tolfenamic acid solubility and SC-CO2 density.
• To optimize model hyperparameters using the Chimp Optimization Algorithm (ChOA).
• To provide reliable predictive tools for pharmaceutical and chemical industries.

Main Methods:

• Utilized a dataset with temperature and pressure as input features.
• Employed three machine learning models: ADA-GPR, ADA-SVR, and ADA-LR.
• Optimized model hyperparameters using the Chimp Optimization Algorithm (ChOA).

Main Results:

• ADA-GPR achieved high accuracy in predicting tolfenamic acid solubility (R-squared: 0.98806) and SC-CO2 density (R-squared: 0.99265).
• ADA-SVR and ADA-LR also showed competitive performance for both predictions.
• The ChOA algorithm effectively enhanced the performance of the machine learning models.

Conclusions:

• Machine learning models, particularly ADA-GPR optimized with ChOA, are effective for predicting tolfenamic acid solubility and SC-CO2 density.
• The developed models offer valuable tools for optimizing processes in the pharmaceutical and chemical sectors.
• This research addresses critical prediction challenges in solubility and density determination.