Development of Python Tools for Unifying EEG Data from Different Devices for Subsequent Mediation and Neural Network Analysis

Abstract

This study addresses the problem of heterogeneity in electroencephalography (EEG) data obtained from different recording devices, which limits the applicability of advanced analytical methods, including mediation analysis and neural network models. A specialized Python-based software tool was developed to unify EEG datasets by standardizing channel ordering, verifying and correcting metadata, and converting data into a consistent format. The approach was tested on recordings obtained from NVX-132 and NeuroAnt encephalographs with different channel configurations and sampling parameters. Spectral characteristics were evaluated across fi ve standard frequency bands. The results demonstrate that the proposed unification procedure preserves the structural and spectral properties of EEG signals. The relative error in spectral power across delta (0.5–4 Hz), theta (4–8 Hz), alpha (8–15 Hz), beta (15–30 Hz), and gamma (30–50 Hz) bands does not exceed 25 %, which is within acceptable limits for EEG analysis. Temporal and phase characteristics of the signals are also maintained, ensuring the reliability of subsequent analyses. The proposed tool enables efficient integration of heterogeneous EEG data and improves the reproducibility and comparability of results in multi-subject and multi-center studies. It facilitates the application of modern machine learning approaches, including convolutional and recurrent neural networks, for the analysis of cognitive processes and clinical diagnostics. The method contributes to the development of standardized pipelines for neurophysiological data processing.