Automated workflow of EIS data validation and quality improvement based on the definition, detection, and removal of outliers

Current EIS validation methods only determine the data quality of a whole impedance spectrum. While EIS data quality is frequently harmed by outliers that are not precisely defined and typically determined by the empirically artificial detection. For the first time, an EIS outlier definition consisting of three qualitative conditions was proposed, which was then taken as a foundation and combined with the improved linear Kramers-Kronig validation, smoothing algorithm, quartiles, and a few sensitive residuals to build an automated workflow to automatically detect and remove outliers. By evaluating the workflow performance on simulated EISs with artificially added disturbances, we selected the most suitable running parameters for the most efficient computation and the highest detection rate of 85.1%, as well as ranked the sensitivities of various residuals to outliers. The results of assessing the workflow on the high-throughput local EISs demonstrated that the workflow outperformed humans in EIS data validation and quality improvement. We made the workflow available to all electrochemists to improve EIS data quality to get more accurate fitting results for process equivalent circuit models.