(342c) Integrating Statistical Analysis and Kinetic Modeling to Increase Learning from High-Throughput Experiments

High-throughput experimentation is widely used in pharmaceutical process development to assess the impacts of process conditions on reaction performance. At BMS, a highly automated workflow is used to execute experimental designs. While statistical analysis performed on the data typically focuses on key responses such as yield, purity, and impurity levels at the reaction endpoint, the automated sampling platforms used to run these reactions also provide a significant amount of kinetic data.

This presentation describes the use of templated analysis scripts to complete statistical analysis of reaction endpoints, plot kinetic data for visualization, and export experimental conditions and data to enable kinetic parameter estimation in DynoChem. The web-based template minimizes the need for manual reformatting of the datasets to accommodate different software platforms, and allows flexibility to customize the analysis to address the requirements of each reaction studied. Case studies highlighting the use of this approach to gain additional insight into reaction performance and to inform process development will be discussed.