Application of a forward design-based multi-attribute decision-making method in quality assessment in pharmaceutical tablet manufacturing

Abstract

With the modernization of the pharmaceutical industry and increasingly stringent regulatory requirements, ensuring the quality and process stability of pharmaceutical tablets has become a critical challenge. In particular, chewable tablets, as a unique type of oral solid dosage form designed to be chewed before swallowing, require special attention to mechanical properties and sensory attributes such as texture, mouthfeel, and taste acceptability. This study aims to implement a forward design-based multi-attribute decision-making (MADM) framework for systematic and accurate multi-criteria quality assessment and prediction of pharmaceutical chewable tablets. Tablets were produced using batch-based tableting technology, and critical quality attributes—including weight variation, friability, porosity, and hardness—were systematically evaluated under varying process conditions. Tensile strength was employed as an external validation index. To overcome the limitations of traditional evaluation methods, the proposed MADM framework incorporates a game theory (GT)-based combinatorial weighting strategy, effectively integrating subjective experience-based judgment with objective data-driven metrics. In addition, a backpropagation neural network (BPNN) model was developed to capture the nonlinear relationships between process parameters and the comprehensive quality index. Compared to conventional weighting approaches such as the analytic hierarchy process (AHP), entropy method (EM), and mean-based weighting, the GT-based strategy demonstrated superior performance in terms of evaluation robustness and accuracy. The MADM results showed strong agreement with the tensile strength data, confirming the reliability and consistency of the evaluation framework. Furthermore, the BPNN model achieved high prediction accuracy, providing solid quantitative support for process optimization. Overall, the proposed method offers a robust and generalizable solution for multi-criteria quality assessment and prediction of pharmaceutical chewable tablets. When integrated with industrial information systems, it enables data-driven optimization, intelligent quality control, and automation, thereby contributing to the advancement of smart pharmaceutical manufacturing.