CSV schema
CSV outputs provide subject-level phenotype tables for downstream analysis. The authoritative documentation layer is the output inventory plus the promoted phenotype pages that name exact output columns.
- Modality
- All modalities
- Pipeline step
- Feature extraction, aggregation, and schema documentation
- Outputs
- Aggregate CSV phenotype tables keyed by eid
- Maturity
- Source-audited reference page
Required schema principles
- Use a stable subject key, usually
eid. - Keep column names stable and documented.
- Preserve units in column names when the pipeline emits them.
- Treat legacy spelling as schema debt, not as an invitation for silent cleanup.
- Validate duplicate
eid, missing columns, empty files, type drift, and column-order drift during aggregation or documentation audit.
Current artifact families
| Artifact family | Registry source | Example rows |
|---|---|---|
| Ventricular volume/function | docs/data/output_column_inventory.yml | LV: V_ED [mL], RV: EF [%], LV: CI [L/min/m^2] |
| Atrial volume/function | output inventory and phenotype dictionary | LA: V_max [mL], RA: EF_total [%], LA: PER-E [mL/s] |
| Myocardial mass/wall thickness | output inventory and phenotype dictionary | Myo: Mass [g], Myo: Thickness (Global) [mm] |
| Mechanics/strain | output inventory and phenotype dictionary | Strain-SAX:*, Strain-LAX:*, Strain-Tagged:* |
| Aortic and flow | output inventory and phenotype dictionary | Aortic Structure-*, Aortic Flow:*, Aortic Distensibility-* |
| Tissue | output inventory and phenotype dictionary | Native T1:*, Native T1-Corrected:* |
| Cross-chamber | output inventory and phenotype dictionary | AV: AVPD [cm], AV: IPVT [mL] |
Schema is part of documentation
CSV column names are not implementation details. They are the bridge between source code, exported data, phenotype pages, validators, and downstream paper tables.
Source audit
- Artifact families were checked against
docs/data/output_column_inventory.yml. - Page-level exact output coverage is enforced for promoted phenotype pages by
website/scripts/validate-phenotype-pages.cjs. - Textbook context boundary: broad clinical textbook context is not surfaced here because this page documents data schema contracts.