Advancing pharmaceutical Intelligence via computationally Prognosticating the in-vitro parameters of fast...

Area of Science:

• Pharmaceutical Sciences
• Computational Chemistry
• Data Science

Background:

• Machine learning (ML) is increasingly used in drug development.
• Tablet efficacy depends on physicochemical properties, formulation, and processing.
• Predicting tablet characteristics aids in optimizing drug delivery.

Purpose of the Study:

• Develop ML models to predict disintegration time, friability, and water absorption ratio for fast-dissolving tablets.
• Evaluate model performance using RMSE and R-squared metrics.
• Provide a novel approach for predicting tablet properties.

Main Methods:

• Data visualization, pre-processing, and splitting.
• Creation and evaluation of ML models including voting regressor, random forest, and KNN.
• Hyperparameter tuning and cross-validation were employed.

Main Results:

• Voting regressor achieved best disintegration time prediction (RMSE: 21.99, R²: 0.76).
• Random forest regressor excelled in friability prediction (RMSE: 0.142, R²: 0.7).
• KNN regressor demonstrated superior performance for water absorption ratio (RMSE: 10.07, R²: 0.94).

Conclusions:

• ML models can accurately predict key tablet properties.
• This approach offers a significant advancement, particularly for friability and water absorption prediction.
• The developed models can streamline tablet development, reducing time and resources.