A hybrid model for multipoint real time potency observation in continuous direct compression manufacturing operations

Area of Science:

• Pharmaceutical Manufacturing
• Chemical Engineering
• Analytical Chemistry

Background:

• Continuous manufacturing is replacing batch processing for oral solid dosage forms.
• Process Analytical Technology (PAT) is crucial for implementing continuous manufacturing.
• Near-Infrared (NIR) spectroscopy offers promising potency measurement but suffers from data noise.

Purpose of the Study:

• To implement an adaptive Kalman filter for robust NIR-derived potency measurement.
• To integrate NIR predictions with a semi-mechanistic compartmental model.
• To enable advanced process control in continuous tablet manufacturing.

Main Methods:

• Developed and implemented an adaptive Kalman filter algorithm.
• Combined NIR potency predictions with a semi-mechanistic compartmental model.
• Utilized in-line measurement data streams for potency inference.

Main Results:

• Achieved more robust concentration estimations.
• Successfully inferred potency levels at multiple locations, including finished tablets.
• Demonstrated the ability to use a single data stream for comprehensive monitoring.

Conclusions:

• The adaptive Kalman filter approach enhances the reliability of NIR-based potency measurements.
• This methodology supports the successful deployment of continuous tablet manufacturing.
• The developed approach facilitates advanced process control strategies for improved pharmaceutical production.