/data-preprocess

Read the neuroflow:phase-data-preprocess skill first. Then follow the neuroflow-core lifecycle: read project_config.md, flow.md, and .neuroflow/data-preprocess/flow.md before starting. Also check .neuroflow/data/flow.md to understand what data is available.

What this command does

Builds and runs a preprocessing pipeline on the ingested data.

Ask:

1. What modality? (EEG, iEEG, fMRI, other)
2. What is the path to the data?
3. What preprocessing steps are needed? Or should it be a standard pipeline for the modality?

Standard EEG pipeline

1. Load data (MNE Raw)
2. Filter (bandpass, notch)
3. Re-reference (average, linked mastoids, or REST)
4. Mark bad channels
5. ICA — fit, inspect components, remove artifacts (EOG, ECG)
6. Interpolate bad channels
7. Epoch around events
8. Baseline correction
9. Reject bad epochs
10. Save preprocessed epochs

Document the pipeline parameters in preprocess-config.md in .neuroflow/data-preprocess/. Write the pipeline script to output_path (from .neuroflow/data-preprocess/flow.md, default: scripts/preprocessing/) — not inside .neuroflow/.

At end

• Save preprocess-report.md — parameters used, bad channels, ICA components removed, epoch rejection rate, subject-level QC notes
• Update .neuroflow/data-preprocess/flow.md
• Append to .neuroflow/sessions/YYYY-MM-DD.md
• Update project_config.md if phase changed