Bridging machine learning and water electrolysis: Concepts, methods, and perspectives

Machine learning (ML) is increasingly applied in water electrolysis research for analysing complex datasets and accelerating catalyst and device development. However, the lack of familiarity with ML concepts and practical workflows among catalysis researchers has limited its effective use. This review provides a concise and practical introduction to ML from the perspective of catalysis researchers. We first outline the basic working principles of ML, including data representation, learning and generalisation. Their working principles are demonstrated in a way accessible to readers without coding experience. Following this, we introduce commonly used software environments and accessible tools, ranging from Python-based workflows to no-code and low-code platforms. Building on this methodological foundation, we systematically review representative ML applications in water electrolysis, organised by key research tasks, including catalyst screening, data-driven analysis of electrochemical and operando measurements, and device-level prediction. Key challenges, such as data consistency, model reliability and physical interpretability, are discussed together with emerging opportunities. In contrast to existing reviews that primarily emphasise algorithmic advances or application case studies, this review seeks to serve as a bridge between ML methodologies and the questions and practices encountered in water electrolysis research.