MSIF-Net: Automated EEG Sleep Staging for REM Sleep Behavior Disorder

A research dataset supporting the validation of the MSIF-Net automated sleep-staging framework for REM sleep behavior disorder (RBD). The model, developed by Huanyu Li, achieved class-wise F1 scores of 0.84 for Wake, 0.94 for NREM, and 0.80 for REM on epoch-level validation. The dataset was last updated on 2026-05-19.

Use Cases

• Validating automated sleep-staging models based on EEG waveforms and time-frequency spectrograms.
• Investigating correlations between automated REM% estimates and clinical symptom scales like PSQI and RBDSQ.
• Benchmarking the performance of multi-stream neural network architectures for polysomnography analysis.

Strengths

• Model performance is quantified with specific F1 scores (0.84 Wake, 0.94 NREM, 0.80 REM).
• Clinical relevance is demonstrated through statistically significant correlations (e.g., ρ = -0.507 with PSQI).
• Released under a permissive CC-BY-4.0 license for open use.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset is small in size (679.5 KB), indicating limited scope.

Provenance

Source

figshare, authored by Huanyu Li.

Collection Method

Data likely originates from polysomnography (PSG) recordings used to train and validate the MSIF-Net model.

Freshness

Last updated 2026-05-19 04:21:05; freshness should be verified.