A Novel Artificial Intelligence System in Formulation Dissolution Prediction.

Area of Science:

• Pharmaceutical Science
• Computational Chemistry
• Drug Discovery

Background:

• Artificial neural network (ANN) techniques are integral to pharmaceutical research for data screening and experimental result prediction.
• Accurate prediction of dissolution results is crucial for efficient drug formulation and development.

Purpose of the Study:

• To develop and validate a novel dissolution result prediction and screening system utilizing a backpropagation network and advanced regression methods.
• To enhance the adaptability of the ANN model for predicting dissolution with varying formulation compositions.

Main Methods:

• A backpropagation artificial neural network (ANN) model was developed and trained using 21 groups of dissolution data.
• Two regression methods, effective data regression method (EDRM) and reference line regression method (RLRM), were employed for prediction.
• A data screening function based on decision tree algorithms was integrated into the system.

Main Results:

• The ANN model demonstrated high accuracy in predicting dissolution results.
• The system requires a smaller database compared to orthogonal experiments for accurate predictions.
• The model successfully adapted to predict dissolution even when formulation composition was altered.

Conclusions:

• The developed ANN model offers a novel, efficient system for drug dissolution prediction and formulation screening.
• This approach significantly decreases the time and cost associated with new drug formulation design.
• The system facilitates easier and more effective design of new pharmaceutical formulations.