Probabilistic Design Space

Abstract In this work we develop a Bayesian framework for building surrogate stochastic models of complex multi-step processes for which tractable mechanistic models are difficult to construct. The probabilistic process envelope defined by the design space provides an extra level of assurance of product quality over and above that provided by traditional process control. While the application we report is specific to drug products manufactured using traditional batch processing, the proposed framework is applicable in general to batch and continuous manufacturing of products that must meet a set of critical product quality specifications.