--- title: 'NSzED: Nigeria Schizophrenia EEG Dataset' url: https://www.emergentmind.com/topics/nigeria-schizophrenia-eeg-dataset-nszed type: topic --- # NSzED: Nigeria Schizophrenia EEG Dataset The Nigeria Schizophrenia EEG Dataset (NSzED) is a publicly available repository of electroencephalogram (EEG) recordings from Nigerian subjects, structured to enable data-driven research in computational psychiatry and neurodiagnostics for schizophrenia. This dataset addresses the underrepresentation of African populations in neuropsychiatric biomarker development, providing a multifaceted resource for the exploration of neural signatures associated with schizophrenia and supporting the development and benchmarking of algorithmic tools for objective diagnosis and prognosis. ## 1. Dataset Structure and Demographics NSzED comprises EEG data from 59 West African subjects of Nigerian origin, including 37 patients clinically diagnosed with schizophrenia and 22 healthy control individuals. All patients were admitted at the Mental Health Ward, Medical Outpatient Department of the Obafemi Awolowo University Teaching Hospital Complex and the Wesley Guild Hospital Unit. Control subjects were medical students and clinicians who volunteered at the same institutions. All participants were characterized using the Mini International Schizophrenia Interview (MINI), with additional assessment using the Positive and Negative Symptoms Scale (PANSS) and the World Health Organization Disability Assessment Schedule (WHODAS) to quantify symptom severity and functional disability, respectively. The sample is predominantly of Yoruba ethnicity, and detailed demographic breakdowns—including age and gender distributions—are provided in the primary dataset publication [2311.18484]. Three recording paradigms are included: - Resting state EEG, - EEG during mental arithmetic tasks (visual and auditory presentation of arithmetic problems), - Auditory task-based EEG (oddball paradigm and 40 Hz Auditory Steady State Responses). ## 2. Data Acquisition Protocols EEG collection used the International 10/20 electrode placement system, ensuring coverage of frontal, central, temporal, parietal, and occipital cortical regions. Two EEG acquisition devices were used: - Contek-2400 system at 200 Hz sampling rate, - BrainAtlas Discovery-24E at 256 Hz. All recordings followed strict impedance constraints (below 5 kΩ) and applied 50 Hz notch filters to mitigate line noise. Auditory tasks employed a sequence of tones in an oddball paradigm, with fixed and flexible cue delivery schemes depending on the acquisition device (synchronized using an open-source lab streaming layer framework). For mental arithmetic tasks, cues were presented either via automated scripts or clinical staff to ensure event synchronization across participants. The following table summarizes the principal acquisition characteristics: | Protocol Aspect | Detail | Source Reference | |--------------------------|-----------------------------------------------------------------------------------------|-------------------| | Subjects (Patients/HC) | 37 / 22 | [2311.18484] | | Electrode montage | 10/20 system | [2311.18484] | | Devices & Sampling Rate | Contek-2400 (200 Hz); BrainAtlas Discovery-24E (256 Hz) | [2311.18484] | | Tasks | Rest, mental arithmetic, oddball auditory, 40 Hz ASSR (Discovery-24E only) | [2311.18484] | | Clinical assessment | MINI, PANSS, WHODAS | [2311.18484] | ## 3. Clinical Curation and Quality Control Participants were rigorously screened prior to inclusion. Patients had to be actively psychotic and/or have prior diagnostic history, as determined by MINI. Control subjects scored baseline-low on PANSS and WHODAS. Quality control steps included the exclusion of recordings from individuals with confounding neurological conditions (including a history of substance use or epilepsy). All datasets underwent further review to ensure signal integrity and remove grossly corrupted or noisy segments before final release. ## 4. Research Applications and Analytical Opportunities The multi-paradigm, multi-metric design of NSzED supports a broad array of research objectives in computational psychiatry: - **Resting state EEG** enables analysis of baseline oscillatory activity, including entropy-based complexity measures (e.g., fuzzy entropy) for assessing dynamical irregularity. - **Task-based EEG** (mental arithmetic) captures cognitive activation, facilitating spatio-temporal pattern analysis and cross-paradigm feature fusion. - **Auditory oddball and 40 Hz ASSR tasks** provide ground for extraction of event-related potentials (ERPs), most notably Mismatch Negativity (MMN) and phase-locked auditory responses. MMN is computed per subject by the difference $\mathrm{MMN}(t) = A_\text{deviant}(t) - A_\text{standard}(t)$ over ERP segments (e.g., 0–450 ms), providing a temporal window into deviance detection mechanisms in schizophrenia. Such multidimensionality is explicitly designed to foster the development of feature fusion techniques, supporting advanced biomarker discovery that integrates information across spatial, temporal, and spectral EEG domains [2311.18484]. ## 5. Challenges and Methodological Constraints NSzED addresses a critical data gap but also introduces several analytical challenges: - Sample size, while substantial for a clinical African EEG cohort, is moderate by international standards, potentially limiting the statistical power of high-dimensional models. - Protocol heterogeneity exists due to the use of two acquisition systems and cue synchronization methods. This may require protocol-specific preprocessing pipelines and harmonization strategies for cross-device data alignment. - The cohort’s ethnic composition (predominantly Yoruba) and regional context may restrict the direct generalizability of extracted biomarkers to other populations. A plausible implication is that rigorous cross-validation, protocol-aware normalization, and demographically matched replication studies will be required for external validation of models developed using NSzED. ## 6. Future Directions and Dataset Evolution NSzED is positioned as an evolving resource—future versions (NSzED-v2 and beyond) are expected to expand subject numbers, broaden ethnic and demographic representation, and incorporate longitudinal EEG recordings to support disease progression and intervention studies. Prospective updates aim to further standardize preprocessing and cue synchronization protocols, increase the range of biomarker modalities (e.g., integrating new ERP measures), and foster international collaborations for comparative computational psychiatry [2311.18484]. The dataset’s accompanying metadata and its curation according to modern open science criteria favor its integration into multi-site studies and benchmarking efforts. Researchers utilizing NSzED are encouraged to build upon the established preprocessing pipelines (baseline correction, band-pass filtering between 1–100 Hz) and explore both hypothesis-driven and data-driven feature extraction strategies. ## 7. Significance for Computational Psychiatry in Low-Resource Settings NSzED constitutes the first richly annotated, multi-task EEG dataset for schizophrenia of African origin. By addressing both clinical and technical heterogeneity, it opens new frontiers for culturally and regionally adapted precision psychiatry, supports the diversification of training corpora for machine learning, and provides a foundation for evaluating the generalizability of computational models developed predominantly on data from non-African cohorts. Its release represents a pivotal advance for neuroinformatics infrastructure on the continent, with substantive implications for the equitable development of EEG-based neurological disorder diagnostics. Source: https://www.emergentmind.com/topics/nigeria-schizophrenia-eeg-dataset-nszed