Infant EEG preprocessing pipelines: A capability framework and current gaps in practice

Infant electroencephalography (EEG) is essential for understanding early brain development, yet presents unique challenges including frequent movement artifacts, low signal-to-noise ratio, and rapid developmental changes. Specialized preprocessing pipelines have been developed to address these issues, but no structured cross-pipeline evaluation of their capabilities currently exists. This article presents a capability framework for ten prominent infant EEG preprocessing tools - nine infant-specific tools (APICE, Adjusted-ADJUST, GADS, NEAR, HAPPE, HAPPE+ER, HAPPILEE, BEAPP, and MADE) and the Modular pipeline, a general-purpose approach scalable to infant data - organized around five capability dimensions: artifact handling, automation and scalability, flexibility and adaptability, developmental sensitivity, and validation against empirical data. The framework is proposed as an exploratory evaluative heuristic, not a definitive taxonomy. For each dimension, each pipeline was assigned a maturity level from 1 (minimal documented support) to 5 (fully demonstrated, well-documented, and developmentally appropriate support), based on predefined anchored descriptors applied independently by two raters (overall Krippendorff's α =.69), with lower agreement on the Automation and Validation dimensions - a pattern consistent with field-level ambiguity about what constitutes strong performance on these dimensions. No single pipeline reached the highest maturity level across all five dimensions; each occupies a distinct profile aligned with particular research needs. Notably, no pipeline achieved level 5 on either developmental sensitivity or validation against empirical data, indicating a field-wide gap in developmental calibration and cross-dataset validation. This capability framework provides a structured foundation for researchers and clinicians to make informed, context-sensitive preprocessing decisions in infant EEG studies.