NMT-4K-EEG Dataset Code and Validation Repository

NMT-4K-EEG Dataset Code and Validation Repository

This is the archived code release for the NMT-4K-EEG Dataset.

This release contains the  code, notebooks, validation scripts, and supporting files used to curate, package, validate, and describe the NMT-4K-EEG dataset. The code was developed to support the preparation of the dataset release and the technical validation reported in the associated Data Descriptor manuscript.

What is included in this repository

• Dataset curation scripts

• Scripts for creating the predefined training and evaluation partitions

• Release packaging scripts

• EDF readability and metadata checking routines

• Channel label and sampling frequency verification code

• Recording duration extraction scripts

• Cross-modal file linkage checks for EEG files, annotation files, and clinical reports

• Event timing validation scripts

• Annotation channel validation routines

• Clinical label normalization code

• SHA-256 checksum generation scripts

• Signal characterization notebooks

• Annotation summary and event distribution notebooks

• Technical validation figure generation notebooks

• Exploratory recording-level baseline analysis code

• Supporting outputs used during dataset inspection and manuscript preparation

Purpose of this release

The purpose of this archived release is to provide a stable and citable version of the code used with the NMT-4K-EEG dataset manuscript. The repository helps document how the dataset was prepared, how the release structure was checked, and how the technical validation outputs were generated.

The code can also be used as a reference for reproducing selected validation checks, inspecting the released file structure, and building similar workflows for clinical EEG dataset validation.

Dataset availability

The NMT-4K-EEG dataset itself is distributed separately through Zenodo under the dataset DOI:

https://doi.org/10.5281/zenodo.21041582

This repository contains the code and validation workflow. It does not contain the full EEG dataset files.

Version note

This Zenodo archive represents the fixed code version associated with the manuscript submission. The GitHub repository may continue to receive minor updates, but this archived release should be used when citing the code version linked to the manuscript.

License

MIT License.