Integrated control strategy based on deep learning and process analytical technology for the manufacture of a triple fixed-dose combination tablet

Abstract

Purpose

This study aims to develop an integrated control strategy using process analytical technology (PAT) and a quality by design (QbD) framework, integrating deep learning for quality prediction to ensure robust manufacturing of a triple fixed-dose combination (FDC) tablet containing chlorthalidone, amlodipine besylate, and valsartan.

Methods

A Box–Behnken design was used to optimize high-shear wet granulation of the chlorthalidone layer and roller compaction of the amlodipine/valsartan layer. Correlations between intermediate quality attributes (QAs) and critical quality attributes (CQAs) were confirmed using Pearson correlation analysis. A deep learning model using artificial neural networks (ANNs) was developed to predict the dissolution profiles with high accuracy. PAT tools have been used for at line monitoring, including 3D imaging system, torque profiles for granule size distribution, and digital image correlation (DIC) for ribbon mechanical properties. A pharmacokinetic study was conducted in healthy human volunteers to assess bioequivalence.

Results

Optimized design spaces defined critical process parameter ranges to ensure the target QAs. Correlation analysis identified the granule size and recompressed granule strength as significant factors affecting dissolution, hardness, and friability. The ANN-based dissolution prediction model exhibited good predictive performance with a low error. An in-process control strategy enables monitoring and process control using PAT tools. The optimized FDC tablet demonstrated bioequivalence with reference products.

Conclusion

This study demonstrates that combining PAT, QbD, and deep learning-based prediction enables real-time control and quality assurance in the manufacturing of complex multidrug oral dosage forms, ensuring consistent product quality and clinical equivalence of the triple FDC tablet.

Continue reading here

Kim, J.Y., Yeo, S., Yoo, JH. et al. Integrated control strategy based on deep learning and process analytical technology for the manufacture of a triple fixed-dose combination tablet. J. Pharm. Investig. (2025). https://doi.org/10.1007/s40005-025-00784-0

Read also our introduction article on Quality by Design (QbD) here: