Contextualizing EEG-Derived Results: A Semantic and Multidimensional Architecture for Cross-Study Analysis

The capture, processing, and analysis of electroencephalographic (EEG) recording data from different experiments, the application of processing methods, and different levels of temporal and spatial granularity are usually stored and analyzed in isolated contexts, making it difficult to trace, compare, and reuse them. This proposal builds upon experience gained from empirical studies involving cognitive assessments, problem-solving tasks, mental imagery experiments, correlation and synchrony analyses, the Hilbert–Huang Transform, and Hurst exponent analysis. These studies reveal a common multidimensional structure that can be leveraged to organize EEG-derived information according to subjects, experiments, methods, channels, brain regions, time scales, and experimental conditions. This work proposes an integrated neuroinformatics architecture for managing, analyzing, and leveraging knowledge derived from EEG data. It integrates a repository of raw EEG data; a semantic layer linking experiments, subjects, methods, processing conditions, and derived results; and a multidimensional environment based on OLAP structures to organize and compare heterogeneous analytical results. Built upon this integrated structure, the proposal includes an AI-supported knowledge management layer designed to identify regularities among accumulated results, contextualize detected patterns using semantic information, and generate candidate knowledge for subsequent validation by specialists. Validated knowledge can be progressively accumulated and reused to interpret new results and support future analyses. The architecture preserves traceability from the generated knowledge back to the analytical results, methods, experiments, and raw EEG data. It also allows for the incorporation of various artificial intelligence and machine learning techniques tailored to the specific characteristics of each analytical problem. The proposed framework offers an extensible foundation for the cross-study analysis of EEG data, reproducibility, and the progressive accumulation of knowledge in neuroinformatics.