Advancing Direct Tablet Compression with AI: A multi-task framework for quality control, batch acceptance, and causal...

Area of Science:

• Pharmaceutical Manufacturing
• Drug Development
• Artificial Intelligence in Pharma

Background:

• Pharma 4.0 is transforming drug development with AI for enhanced efficiency.
• Direct Tablet Compression (DTC) relies heavily on API and excipient properties.
• Optimizing tablet quality requires understanding complex interactions.

Purpose of the Study:

• To develop a novel multi-task AI framework for predicting tablet properties.
• To determine batch acceptance and provide insights for quality improvement.
• To enable real-time monitoring and regulatory compliance in pharmaceutical manufacturing.

Main Methods:

• A multi-task framework combining regression, classification, and text generation.
• Integration of statistical methods, Neural Networks (NN), Natural Language Processing (NLP), and Generative AI (GenAI).
• Validation using the Harvard Dataverse V1 dataset of Fast Disintegrating Tablets (FDTs).

Main Results:

• Achieved 91.8% R² for regression and 95.5% accuracy for classification.
• Demonstrated superior performance over state-of-the-art methods.
• Successfully predicted tablet properties (friability, hardness, disintegration, water absorption).

Conclusions:

• The AI framework significantly enhances pharmaceutical manufacturing quality control.
• It provides actionable insights for optimizing tablet production and identifying root causes of defects.
• The approach supports real-time monitoring, compliance, and improved drug development outcomes.