(419h) Experimental Design: From Iterative Reaction Optimization to Route Planning with Literature Data

Experimental design techniques are a quintessential part of the engineer’s toolbox. From its origins in statistical DOE to sequential experimentation, these algorithms have been applied to many reaction optimization problems with great success. The goal at each stage of decision-making is to select the ‘best’ experiment given information from previous experiments. This concept can be generalized to the broader challenge of selecting the ‘best’ experiment given all prior information, including information from the literature. In this talk, I will discuss the path from robust statistical methods to more recent work in computer-aided synthesis planning that incorporates cheminformatics and machine learning techniques. Through this lens, data-driven tools trained on millions of reactions from the literature provide a prior that is refined and updated by new experimental screening.