Development of an AI-Based Predictive Quality Control Model to Minimize Batch Variability in Tablet Manufacturing

Abstract

A long standing issue that has been faced in the manufacturing of the tablets is a batch to batch impairment which has resulted in the deviation of the product quality, higher costs and regulatory issues. Conventional quality control methods are mainly reactive where end products are tested and this restricts the possibility of having defects prevented prior to the actual production process. In this work, it is recommended to use an AI-assisted predictive quality control (PQC) model that reduces the variability of batches using historical process information, material characteristics that are critical, and process critical parameters. A number of machine learning models, such as the Random Forest, Support Vector machine, XGBoost, and Artificial neural networks, were tested to present the outcomes of machine learning in predicting the quality of tablets, including hardness, friability, change in weight, and dissolution. The findings were that XGBoost was better than other models because it had the highest predictive accuracy (R2 = 0.92, RMSE = 1.97), and it ablely represented nonlinear relationships between the process variables. The analysis of the importance of features has determined compression force, granulation moisture content, and API particle size as the most important factors affecting the quality of the batches. It is possible to predict the at-risk batches in advance, make real-time decisions, and lower the costs of reprocessing with the help of the proposed predictive QC model and adhere to the principles of Quality 4.0 and Industry 4.0. The study offers a new paradigm of AI implementation to drug production and offers real-life tips on how it can be used to make the processes more efficient and to achieve consistent quality of the product and minimized batch failures.

Keywords:

Predictive Quality Control, Tablet Manufacturing, Batch Variability, Artificial Intelligence, Machine Learning, Industry 4.0, Quality 4.0