Bayesian Optimal Experimental Design (BOED) provides a rigorous framework for selecting experiments that maximise expected information gain. Despite its theoretical appeal, BOED remains underutilised in applied sciences like chemistry, largely due to the significant computational demands it imposes.
In this talk, I present a computationally efficient BOED framework tailored to chemical reaction kinetics. By combining Gaussian process emulation with Bayesian design principles, we reduce the cost of evaluating complex utility functions, making the design of informative experiments both tractable and practical. A key focus is on optimising control parameters to enhance the precision of kinetic parameter estimates, particularly in data-sparse experimental regimes where classical approaches struggle.
This work sits at the interface of statistics, machine learning, and the physical sciences, and aims to make principled design methods more accessible to practitioners. I will discuss both methodological and applied aspects, including the integration of prior information, uncertainty quantification, and broader implications for experiment planning in scientific domains.