A comparative simulation study of cluster ensemble algorithms integrated with multiple imputation for clustering with missing data.

Electrical Status Epilepticus during Sleep (ESES) is an age-dependent epileptic encephalopathy characterized by significant sleep-induced epileptiform discharges (spike-wave index ≥ 85%) and acquired cognitive or behavioral deficits [–]. As a rare but impactful syndrome, ESES poses a major threat to neurodevelopment, with prolonged activity often leading to enduring impairments in language, memory, and executive functions, even after seizure remission [–]. Consequently, elucidating the neural mechanisms underlying cognitive dysfunction in ESES is critical for advancing early diagnosis and targeted interventions. While traditional neuropsychological scales, such as the Wechsler Intelligence Scale for Children (WISC), provide valuable assessments of cognitive domains and are recognized for their comprehensiveness and reliability [, ], they are limited by subjectivity and an inherent inability to detect deficits before they become clinically overt []. There is, therefore, an urgent need for objective biomarkers that can precisely quantify neural alterations associated with ESES-related cognitive impairment. Resting-state functional magnetic resonance imaging (rs-fMRI) offers a powerful, non-invasive window into brain function, enabling the investigation of neural circuits and cognitive processes [–]. Among its analytical methods, Regional Homogeneity (ReHo) is particularly suited for investigating local neural synchrony and has been widely used to identify localized brain dysfunction in various neurological and psychiatric disorders [–]. By measuring the temporal synchronization of neural activity within a given brain region, ReHo can reveal subtle alterations in local brain function that may serve as sensitive biomarkers.