Merging Real-Time NIR and Process Parameter Measurements in a Fluidized Bed Granulation Process to Predict Particle...

Area of Science:

• Pharmaceutical Manufacturing
• Process Analytical Technology (PAT)
• Particle Engineering

Background:

• Controlling critical quality attributes like granule moisture and particle size is vital for drug product quality.
• Fluidized bed granulation (FBG) is a key process for preparing powders for oral solid dosage forms.
• Accurate particle size determination is essential for powder flow and subsequent processing.

Purpose of the Study:

• To develop an improved inline model for predicting particle sizes in FBG.
• To combine near-infrared spectroscopy (NIR) with process parameters for enhanced prediction accuracy.
• To validate the model's performance across different batch datasets.

Main Methods:

• Development of a 'merged-partial least square (PLS)' model integrating NIR data and FBG process parameters.
• Prediction of particle size distribution (Dv10, Dv25, Dv50, Dv75, Dv90) at the process endpoint.
• Model validation using training and testing datasets from multiple batch runs.

Main Results:

• The merged-PLS model demonstrated improved prediction accuracy (lower RMSEP) compared to NIR-only or process parameters alone models.
• Enhanced root-mean-squared error of prediction (RMSEP) was consistently observed across all particle size fractions.
• Validation with different batch splits confirmed the model's robust predictive performance.

Conclusions:

• The merged-PLS model significantly enhances the prediction of endpoint particle sizes in FBG.
• This improved prediction capability aids in better process understanding and control.
• The developed method offers a valuable tool for optimizing the FBG process in pharmaceutical manufacturing.