An open-access EEG dataset from indigenous African populations for schizophrenia researchZenodo
Machine-learning pipelines for schizophrenia demand large, ethnically diverse electroencephalography (EEG) corpora, yet African populations remain under-represented in the public domain. The African Schizophrenia EEG Dataset (ASZED-153) helps close this gap with 153 raw, 16-channel recordings from 7...
Saved in:
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-10-01
|
| Series: | Data in Brief |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2352340925006584 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Machine-learning pipelines for schizophrenia demand large, ethnically diverse electroencephalography (EEG) corpora, yet African populations remain under-represented in the public domain. The African Schizophrenia EEG Dataset (ASZED-153) helps close this gap with 153 raw, 16-channel recordings from 76 clinically characterized patients and 77 matched controls recruited in south-western Nigeria (mean age ≈ 39 years). Signals were acquired at two hospital units using Contec KT-2400 (200 Hz) and BrainMaster Discovery24-E (256 Hz) systems under harmonized protocols, retaining only the devices’ default filter settings.Each session contains four paradigms—eyes-closed resting state, arithmetic working-memory, auditory oddball to elicit mismatch negativity, and a 40 Hz auditory steady-state response—so oscillatory, ERP and cognitive-load markers can be compared within the same individuals. Recordings are released unchanged in European Data Format, accompanied by structured .gnr sidecars detailing clinical scores, device settings and protocol metadata, enabling transparent end-to-end pipelinesData are organized in a version-controlled tree with a public key-map, allowing new African centers to append recordings without breaking existing scripts and paving the way for federated growth beyond Nigeria. By uniting ancestral diversity, multi-task paradigms and minimal preprocessing, ASZED-153 will allow researchers audit ancestry-linked performance drift in existing classifiers, probe biomarkers that may be masked in Euro-Asian cohorts, benchmark algorithms across hardware heterogeneity, and prototype reproducible, open science workflows.ASZED-153 is openly available via Zenodo under a CC-BY licence, and contributions to future releases are welcomed. We anticipate that this resource will accelerate the development of fair, generalizable and clinically useful EEG-based tools for schizophrenia worldwide. |
|---|---|
| ISSN: | 2352-3409 |