Modern product formulation has never been more complex. Whether developing a new shampoo or a life-saving drug, today's scientists and engineers face a new reality: ingredients are no longer simple. They come with sub-choices. Each ingredient may have multiple sub-ingredient options, each with its own concentration limits, and each formulation may require selecting only a handful from a longer list of candidates. Defining a valid experimental space under these layered constraints is a new challenge, which means that the better-known approaches are no longer sufficient.
In this presentation, we introduce a new JMP add-in developed to meet exactly this challenge. The add-in gives researchers an intuitive interface to define complex mixture structures, including hierarchical ingredient categories, upper and lower concentration bounds, and limits on how many sub-ingredients can be active in any given formulation. From these inputs, a candidate set table is constructed to be fed to the JMP Bayesian Optimization platform.
We then bring this to life with a real-world application, where the add-in was used to design experiments for a new formulation, navigate a highly constrained ingredient space, and ultimately identify a winning formula faster and with greater confidence than traditional approaches would allow.
Presenters
Skill level
- Beginner
- Intermediate
- Advanced
