Adaptive Optimization of Chemical Reactions with Minimal Experimental Information

Abstract

Optimizing reaction conditions is cornerstone in synthetic chemistry and connected to expert chemistry knowledge, as well as laborious exploration of reaction parameters. To automate reaction condition optimizations and augment chemical intuition, we developed adaptive machine intelligence to navigate search spaces. Our approach (LabMate.AI) employs an interpretable algorithm and requires only 0.03–0.04% of all search space as input data. LabMate.AI optimizes many reaction parameters simultaneously, with minimal computational resources and time. In proof-of-concept studies with distinct goals, we demonstrate how LabMate.AI can identify optimal conditions for an Ugi and a C–N cross-coupling reaction in a competitive manner relative to human experts without requiring specialized chemistry equipment and using orders of magnitude less data than related approaches. LabMate.AI affords quantitative and interpretable reactivity insights, is rapidly deployable and autonomously formalizes chemical intuition, thereby providing an innovative framework towards informed and automated experiment optimization and democratization of synthetic chemistry.