(340n) Development of Condition Monitoring Systems to Support Continuous Manufacturing of Pharmaceutical Oral Solid Dosages.

The main focus of this research is to support Continuous Manufacturing, which is hailed as the future of Pharmaceutical Manufacturing, but relies heavily on the minimization of interruptions (due to abnormal operating conditions). Faults that eventually lead to failure need to be detected and intervened promptly, and this can be achieved through condition monitoring. Hence, my research is focused on establishing a framework for developing a condition monitoring system for the Continuous Manufacturing of Pharmaceutical Oral Solid Dosages (such as tablets), albeit the research is meant to cover all aspects of drug manufacturing. Developing these condition monitoring systems require the deployment of novel sensors (e.g. NIR and Direct Camera Imaging) that can infer the Critical-to-Quality Attributes (CQA) of the process in real-time, and the creation of condition monitoring models that consider four fault types (i.e., whether related to the material, sensor, controller, or equipment). Machine learning is then employed to establish the relationship between the sensor data and the CQAs, and to create the condition monitoring models, which essentially maps the process data to the “Condition” of the process (i.e., whether the process is under normal operating condition, or not). Two machine learning approaches (i.e. traditional and model-based machine learning) are being considered in the research effort, and these are utilized accordingly based on scarcity of data, and the modular nature of the components of a pharmaceutical manufacturing process.