This project addresses a key challenge faced by both manufacturers and regulators: assessing whether the best data analytical tools are utilized for risk-based decision-making. We are developing and validating decision support tools which can be used to determine the appropriate process models and can quantitate the degree of risk associated with use of the selected model for process decision-making. We are developing tools that help manufacturers select the appropriate mathematical models to describe their processes and upon which to base their control strategies. Manufacturers can use these tools to quantitatively assess the process risks associated with their choice of manufacturing model. We are validating these tools experimentally in a continuous monoclonal antibody manufacturing testbed that features state-of-the art process equipment and in-line and at-line analytics. This project also applies advanced data analytics such as machine learning, natural language processing, and artificial intelligence to incorporate all of the data sources available within the manufacturing plant and the biomedical ecosystem to develop tools to optimize manufacturing operations and to use all of the data available to ensure the reliable supply of safe and effective biologic medicines.
- FDA Center for Drug Evaluation & Research
- Weike Sun and Richard D. Braatz. Opportunities in tensorial data analysis for chemical and biological manufacturing processes. FOPAM Special Issue, Computers and Chemical Engineering, accepted on September 12, 2020.
- Weike Sun and Richard D. Braatz. ALVEN: Algebraic Learning Via Elastic Net for static and dynamic nonlinear model prediction. FOPAM Special Issue, Computers and Chemical Engineering, accepted on September 12, 2020. Galley proof in production.