Extract Data Table
PREREQUISITE: This skill needs an initialized jinko-sdk connection and an
SDK satisfying its metadata.requires_sdk range. Run the jinko-sdk-setup skill
(../jinko-sdk-setup/SKILL.md) and proceed only once its check passes. If that
skill is not found, install it from novainsilico/jinko-skills.
Preserve what the source reports. Keep estimated, transformed, and directly
transcribed values distinguishable.
Inputs
Require the source artifact and the requested series. For a Jinkō-ready output,
also require the target obsId mapping, model units, and any scenario/arm scope.
Ask for missing mappings rather than guessing them.
Workflow
- Locate each requested series and record its citation plus page, table, figure,
panel, or supplement. Prefer machine-readable tables over OCR and OCR over
graphical digitization. If no suitable extraction tool is available, request
a tabular source instead of estimating visually.
- Extract only reported values. For graphical digitization, retain the raw
digitized points and identify them as estimates. Do not fit, smooth, aggregate,
or impute unless explicitly requested; record any such transformation.
- Preserve the reported statistic. Do not interchange raw values, means,
medians, SD, SE, confidence intervals, IQR, or min/max. Convert units only when
the source and target units are known, recording the formula and original
values. Express Jinkō time values as ISO-8601 durations.
- For general extraction, emit a readable CSV or Markdown table with series,
time/condition, value or bounds, unit, and source locator.
- For Jinkō output, follow the row schema owned by
jinko-data-table. Use point
rows for reported point values and range rows only for reported lower/upper
bounds. Set obsId, armScope, unit, and experimentRef explicitly.
- Run the
jinko-data-table creation script in dry-run mode with the expected
--allowed-obs-id values, --require-unit, and --require-experiment-ref.
On approval, add --require-fitness --apply. The script owns row/schema
checks, upload, and the server validForFitnessFunction gate.
Return
Return the extracted file and a compact report containing:
- source locator and extraction method for each series;
- original statistic, units, and any transformations or conversions;
- assumptions, unreadable values, and digitization uncertainty;
- for Jinkō output, data-table SID, URL, observable mapping, and confirmed
validForFitnessFunction: True.
If the server does not explicitly report fitness compatibility as true, return
the table as not calibration-ready. Never silently replace or manufacture values
to make a table pass validation.
1---2name: jinko-task-extract-data-table3description: Extract or digitize reported biomedical values from papers, figures, tables, supplements, images, or web sources into traceable CSV/Markdown, optionally as a calibration-ready Jinkō data table. Use when numeric evidence must be transcribed, normalized, unit-converted, or bound to model observables. Do not use for literature discovery, evidence synthesis, or inventing values absent from the source.4license: MIT5---67# Extract Data Table89> **PREREQUISITE:** This skill needs an initialized `jinko-sdk` connection and an10> SDK satisfying its `metadata.requires_sdk` range. Run the `jinko-sdk-setup` skill11> (`../jinko-sdk-setup/SKILL.md`) and proceed only once its check passes. If that12> skill is not found, install it from `novainsilico/jinko-skills`.1314Preserve what the source reports. Keep estimated, transformed, and directly15transcribed values distinguishable.1617## Inputs1819Require the source artifact and the requested series. For a Jinkō-ready output,20also require the target `obsId` mapping, model units, and any scenario/arm scope.21Ask for missing mappings rather than guessing them.2223## Workflow24251. Locate each requested series and record its citation plus page, table, figure,26 panel, or supplement. Prefer machine-readable tables over OCR and OCR over27 graphical digitization. If no suitable extraction tool is available, request28 a tabular source instead of estimating visually.292. Extract only reported values. For graphical digitization, retain the raw30 digitized points and identify them as estimates. Do not fit, smooth, aggregate,31 or impute unless explicitly requested; record any such transformation.323. Preserve the reported statistic. Do not interchange raw values, means,33 medians, SD, SE, confidence intervals, IQR, or min/max. Convert units only when34 the source and target units are known, recording the formula and original35 values. Express Jinkō time values as ISO-8601 durations.364. For general extraction, emit a readable CSV or Markdown table with series,37 time/condition, value or bounds, unit, and source locator.385. For Jinkō output, follow the row schema owned by `jinko-data-table`. Use point39 rows for reported point values and range rows only for reported lower/upper40 bounds. Set `obsId`, `armScope`, `unit`, and `experimentRef` explicitly.416. Run the `jinko-data-table` creation script in dry-run mode with the expected42 `--allowed-obs-id` values, `--require-unit`, and `--require-experiment-ref`.43 On approval, add `--require-fitness --apply`. The script owns row/schema44 checks, upload, and the server `validForFitnessFunction` gate.4546## Return4748Return the extracted file and a compact report containing:4950- source locator and extraction method for each series;51- original statistic, units, and any transformations or conversions;52- assumptions, unreadable values, and digitization uncertainty;53- for Jinkō output, data-table SID, URL, observable mapping, and confirmed54 `validForFitnessFunction: True`.5556If the server does not explicitly report fitness compatibility as true, return57the table as not calibration-ready. Never silently replace or manufacture values58to make a table pass validation.
novainsilico