Figure 1.

Overview of the design of research and main steps of analysis. Two independent EEG datasets were included in this research, dataset 1 was used for uncovering the neuropathic mechanism of PKD and training a prediction model for the classification of different PKD remission states, while dataset 2 was used as testing data for validation of the CPM. High‐density EEG was recorded under eye‐open (EO) and eye‐close (EC) paradigms. The activity of cortical and subcortical regions was estimated by source localization. Features of regions’ power functional connectivity (PEC) between two regions in six oscillations were calculated. Neural activity patterns of PKD were investigated by comparing features between HC and PKD or HC with different PKD subtypes. A CPM model was established based on features selected from dataset 1 and validated by dataset 2.