Subscribe Bookmark RSS Feed



Aug 19, 2014

Explore and Exploit: Extending the Art of Model Generation, Evaluation and Visualization


Martin Owen, Quality by Design Development Leader, GlaxoSmithKline martin.owen
Don Clancy, Scientific Investigator, GlaxoSmithKline
Neil Hodnett, Scientific Investigator, GlaxoSmithKline
Rachel Orr, Accelerated Stability Leader, GlaxoSmithKline
David Burnham, Consultant, PegaAnalytics Dave@Pega-Analytics

Stability of medicines is key to the safety of the patient. Medicines are commonly packaged in a bottle or blister pack. Over time, moisture permeates the packaging, which can cause instability in the active drug substance, resulting in increased impurity levels and decreased efficacy. A JSL application has been written to model the storage conditions and to predict moisture intake over time. In conjunction with this, design of experiments is used to accelerate the impact of moisture and temperature on the physical properties of the medicine. Using the JMP Nonlinear platform, kinetic models are built to project shelf-life properties of the medicine. Traditionally this modeling has been time-consuming and relied on domain experts. The JSL application automatically evaluates competing models and applies heuristic rules to facilitate model selection – choosing models based on statistical properties, plus meaningful scientific interpretation. The application handles the associated data-handling workflows and delivers a capability for engaging multidisciplinary problem solvers. All information is presented visually and interactively in a dashboard format. This enables decision makers, who may not themselves build the models, to access advanced analytical and statistical methods and explore what-if scenarios in real time and evaluate that the quality of the model is fit for purpose.