The pharmaceutical industry is undergoing a paradigm shift from traditional manufacturing to AI-powered manufacturing. However, there are different challenges and complexities that need to be overcome for an efficient and successful implementation of AI technology in the industry. For example, currently, AI/ML model development programs are scattered and difficult to use, specifically for those scientists who have less experience in computer programming (e.g. Python). Different computer programs are needed for the deployment of different AI methods. The available AI programs are also not suitable for easily connecting with manufacturing processes, limiting their applications only to model development and not to their deployment for real time applications. There are no AI/ML programs dedicated to pharmaceutical manufacturing where achieving consistent product quality is very crucial. In order to widely apply AI/ML tools within Pharmaceutical manufacturing, the emerging AI platform needs to be reliable, simple, user friendly, and should be capable of supporting the achievement of high product quality. This work is precisely focused on overcoming these challenges and complexities and identifying the correct way of utilizing AI technology to improve and regulate product quality efficiently and effectively.
The main goal of this work is to develop an ‘advanced AI/ML toolbox’ needed for continuous pharmaceutical manufacturing processes. Such a toolbox will be useful for process digitalization, real time product quality assurance, digital regulatory evaluation of drug substance quality, and to handle other complexities associated with advanced continuous manufacturing. The toolbox consists of different AI/ML methods such as Artificial Neural Network (ANN), Dense Neural Network (DNN), Convolution Neural Network (CNN), Long short-term memory (LSTM), and Generative Adversarial Networks (GANs).
In this work, four machine learning (ML) models have been trained to predict the response of continuous pharmaceutical manufacturing process and the performance of these ML models has been compared. The investigated ML methods are long short-term memory (LSTM), convolution neural network (CNN), random forest (RF), and artificial neural network (ANN). The best performing ML model is then selected for implementation into the continuous pharmaceutical manufacturing process for real time prediction. Other methods and tools needed for pharma 5.0 namely modelling, material traceability, optimization, advanced control, cyber-physical security, and data management have been developed for the continuous pharmaceutical manufacturing process. A systematic framework including the methods and tools have been developed for material traceability. A corresponding software tool has been also developed to automate the material traceability procedure. A method has been developed for dynamic optimization of the feeder refill strategies. An advanced model predictive control (MPC) system has been developed and implemented in the continuous pharmaceutical manufacturing (CPM) pilot-plant. The CPP’s and CQA’s are controlled in real time. An RTD tool box has been developed to control the CQA’s in real time. A novel software tool named CPS (Cyber-Physical Security) has been developed for cyber-physical security of the continuous pharmaceutical manufacturing.
The objective of this presentation is two-fold; first to highlight the developed methods and tools and then demonstrate its application for continuous pharmaceutical manufacturing process.