PDF(Pubmed)
Abstract：
In this review, recent advances regarding the integration of machine learning into electrochemical analysis are overviewed, focusing on the strategies to increase the analytical context of electrochemical data for enhanced machine learning applications. While information-rich electrochemical data offer great potential for machine learning applications, limitations arise when sensors struggle to identify or quantitatively detect target substances in a complex matrix of non-target substances. Advanced machine learning techniques are crucial, but equally important is the development of methods to ensure that electrochemical systems can generate data with reasonable variations across different targets or the different concentrations of a single target. We discuss five strategies developed for building such electrochemical systems, employed in the steps of preparing sensing electrodes, recording signals, and analyzing data. In addition, we explore approaches for acquiring and augmenting the datasets used to train and validate machine learning models. Through these insights, we aim to inspire researchers to fully leverage the potential of machine learning in electroanalytical science.
摘要：
在这次审查中,概述了将机器学习集成到电化学分析中的最新进展，专注于增加电化学数据分析背景的策略，以增强机器学习应用。虽然信息丰富的电化学数据为机器学习应用提供了巨大的潜力，当传感器难以识别或定量检测复杂的非目标物质基质中的目标物质时，就会出现局限性。先进的机器学习技术至关重要，但同样重要的是方法的发展，以确保电化学系统可以产生数据与不同的目标或单一目标的不同浓度的合理变化。我们讨论了为构建这种电化学系统而开发的五种策略，在制备感测电极的步骤中采用，记录信号，和分析数据。此外,我们探索获取和增强用于训练和验证机器学习模型的数据集的方法。通过这些见解,我们的目标是激励研究人员充分利用机器学习在电分析科学中的潜力。
