ASZED, African Schizophrenia EEG Dataset

Description

ASZED — the African Schizophrenia EEG Dataset — is a clinical scalp EEG corpus described by its authors as “the first publicly available EEG dataset from African indigenous populations for schizophrenia studies” (Mosaku et al., 2024). It comprises 153 raw recordings from 76 clinically characterized schizophrenia patients and 77 matched healthy controls recruited at two hospital units in south-western Nigeria, with each session spanning four paradigms (resting state, an arithmetic/working-memory task, an auditory oddball for mismatch negativity, and a 40 Hz auditory steady-state response) so that oscillatory, ERP, and cognitive-load markers can be compared within the same individual (Olateju et al., 2025). The motivation is representational: machine-learning pipelines for schizophrenia demand large, ethnically diverse EEG corpora, yet African populations remain under-represented in public neuroimaging data. The dataset was collected by a team led by K.S. Mosaku at Obafemi Awolowo University (with collaborators at the University of Ilorin) and grew out of an earlier pilot release, the Nigerian Schizophrenia EEG Dataset (NSzED) (Olubadewo-Joseph et al., 2023).

Citation

• Paper DOI: TODO(confirm)
• Dataset DOI: 10.5281/zenodo.14178398
• Reference: Olateju, E.O., Mosaku, K.S., Ayodele, K.P., et al. (2025). An open-access EEG dataset from indigenous African populations for schizophrenia research. Data in Brief (ScienceDirect PII S2352340925006584). Dataset: Mosaku, K.S., et al. (2024). ASZED — The African Schizophrenia EEG Dataset (v1). Zenodo.

Catalog citation block

• Olateju, E.O., Mosaku, K.S., Ayodele, K.P., Akinsulore, A., Ajiboye, P.O., Ayorinde, A., Agboola, O., Obayiuwana, E., Akinwale, O.B., & Oyekunle, W.A. (2025). An open-access EEG dataset from indigenous African populations for schizophrenia research. Data in Brief (Elsevier). ScienceDirect PII S2352340925006584. https://www.sciencedirect.com/science/article/pii/S2352340925006584
• Mosaku, K.S., Olateju, E.O., Ayodele, K.P., et al. (2024). ASZED — The African Schizophrenia EEG Dataset (Version v1) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.14178398
• Olubadewo-Joseph, A., et al. (2023). Nigerian Schizophrenia EEG Dataset (NSzED) Towards Data-Driven Psychiatry in Africa. arXiv:2311.18484. https://arxiv.org/abs/2311.18484

Download and access

Field	Value
Official source	https://zenodo.org/records/14178398
Access class	open
Size	ASZED-153.zip, ~207.7 MB compressed (Zenodo)
Format	edf, gnr, kmp
BIDS	false
Mirrors	none recorded

Source & access

• Repository: Zenodo, record 14178398 (concept/parent DOI 10.5281/zenodo.14178397; version v1 DOI 10.5281/zenodo.14178398), published 2024-11-18.
• DOI: 10.5281/zenodo.14178398
• License: Creative Commons Attribution 4.0 International (CC-BY-4.0).
• Data-descriptor paper: Data in Brief (Elsevier), open access — An open-access EEG dataset from indigenous African populations for schizophrenia research (ScienceDirect PII S2352340925006584, 2025).
• Archive: distributed as ASZED-153.zip (~207.7 MB compressed on Zenodo).

Data structure

The corpus is distributed as raw, device-native files rather than a BIDS tree. The .gnr and .kmp extensions are BrainMaster/Discovery acquisition-native files (raw recording + montage/key-map), consistent with the dual-amplifier acquisition — the BrainMaster arm appears to contribute the .gnr/.kmp files while sessions were also exported to EDF. No derivatives or preprocessing pipeline are bundled; the data are raw. No datalad/git-annex packaging is used (plain Zenodo zip).

License

• Name: CC-BY-4.0
• Snippets on this site: allowed — the site may ship short derived snippets and precomputed products from this dataset, labelled as derivatives (re-referenced, filtered, cropped) with citation, license and DOI.

Acquisition

Field	Value
Device	Contec KT-2400 (200 Hz) and BrainMaster Discovery24-E (256 Hz); 10-20 montage
Device class	research-cap
Channels	16
Sampling rate (Hz)	200 (registry value: 200/256)
Online filters	each device’s default filter settings (NSzED precursor applied a 50 Hz notch)
Reference	not fully documented; device-dependent
Mains frequency (Hz)	50
Paradigms	rest-ec, mental-arithmetic, mmn-oddball, assr-40hz
Participants	153
Population	76 schizophrenia patients + 77 controls, Nigeria; likely age/sex imbalance between groups
Clinical groups	schizophrenia, HC
Sessions	1
Durations	per-task recording durations not documented in the open metadata

Acquisition

Signals were acquired at two hospital units under harmonized protocols using two different amplifiers, retaining only each device’s default filter settings (Olateju et al., 2025):

• Contec KT-2400 at 200 Hz sampling, and
• BrainMaster Discovery24-E at 256 Hz sampling.

Recordings use the international 10/20 system. The released ASZED-153 is documented as 16-channel; the NSzED precursor described an 18-electrode symmetric montage (Fp1, Fp2, F3, F4, F7, F8, C3, C4, Cz, T3, T4, T5, T6, P3, P4, Pz, O1, O2) with impedances kept below 5 kΩ and a 50 Hz notch applied, so the exact channel set in the public 16-channel release should be confirmed against the EDF headers (Olubadewo-Joseph et al., 2023). Reference/ground details and per-task recording durations are not fully documented in the open metadata; treat them as device-dependent.

Participants

The release (“ASZED-153”) contains 153 subjects: 76 schizophrenia patients and 77 matched healthy controls, with an overall mean age of approximately 39 years (Olateju et al., 2025). Per-group age and sex breakdowns for the full release are not itemized in the open metadata. The precursor NSzED pilot reported a smaller subset (37 patients, 22 controls) with patients meaningfully older than controls (patients ~40.2 ± 12.8 years, range 20—74; controls ~29.0 ± 7.7 years, range 20—42) and a sex imbalance (patients 20F/17M; controls 17F/5M), so a similar age/sex confound between groups should be expected in the full release until verified against per-subject metadata (Olubadewo-Joseph et al., 2023). Patients were clinically characterized using the MINI, PANSS, and WHODAS instruments; recruitment was through the Obafemi Awolowo University Teaching Hospital Complex (Ile-Ife and Ilesa), Nigeria.

Tasks / conditions

Each session contains four paradigms (Olateju et al., 2025):

1. Eyes-closed resting state — baseline oscillatory activity.
2. Arithmetic / working-memory task — cognitive activation.
3. Auditory oddball (mismatch negativity, MMN) — in the NSzED protocol, standard tones 1 kHz/100 ms with frequency- and duration-deviants (3 kHz/100 ms and 1 kHz/200 ms).
4. 40 Hz auditory steady-state response (ASSR) — gamma-range entrainment, a well-known schizophrenia biomarker.

Used on this site

L4.5, L7.6

Loader

Fetches only the stated subset. Recorded loader: HTTPS (Zenodo) — see snippet.

# Zenodo record 14178398 distributes the whole release as one archive (~207.7 MB).
wget https://zenodo.org/records/14178398/files/ASZED-153.zip
unzip ASZED-153.zip

Caveats worth knowing

• Two amplifiers at 200 and 256 Hz with device-default filters; verify the channel set from EDF headers.

Notable caveats

• Heterogeneous acquisition. Two amplifiers at two sites with different sampling rates (200 vs 256 Hz) and device-default filters — resampling and explicit filtering are required before pooling, and any spectral analysis must account for the rate difference.
• Line noise. Nigeria uses 50 Hz mains; the NSzED protocol applied a 50 Hz notch, so expect 50 Hz (not 60 Hz) line artifacts and harmonics.
• Channel-count discrepancy. The public release is described as 16-channel while the precursor documented 18 electrodes — verify the actual montage from EDF headers before assuming a fixed layout.
• Group confounds. Patients are likely older than controls and sex distributions differ (per the NSzED subset); age and sex should be modeled as covariates in any patient-vs-control contrast.
• Mixed file formats. EDF plus BrainMaster-native .gnr/.kmp — a loader must handle both, and the EDF and .gnr sets may not be a clean one-to-one mapping across subjects/tasks.
• N caveat. “ASZED-153” denotes 153 recordings/subjects; do not conflate with the smaller NSzED pilot counts (37 + 22).

Related datasets

• ERP CORE (Compendium of Open Resources and Experiments)
• EEG Motor Movement/Imagery Dataset (EEGMMIDB)
• LEMON (MPI-Leipzig Mind-Brain-Body)
• Dortmund Vital Study resting EEG
• SRM Resting-state EEG

# Loader recorded in the catalog: HTTPS (Zenodo) — see snippet
# TODO(confirm) the exact call and the subset to fetch. Official source: https://zenodo.org/records/14178398