Structured data extraction

From PDF to analysis-ready data—with the source in view.

Extract quantitative evidence, verify it against the article, keep reviewer drafts independent and reconcile only after the evidence is checked.

✓ Reviewer drafts✓ Source verification✓ Statistical routing

Extracted evidenceSource verified

Outcome	Value	Source	Route
Mortality	14 / 208	Table 2	Binary
Length of stay	8.4 ± 2.1	Table 3	Continuous
Overall survival	HR 1.42	Table 4	HR
Myosteatosis	OR 1.78	Fig. 2	Verify

Source-first rule: confirm group, outcome, unit, time point and reference category before moving evidence into a dataset.

Extraction safeguards

Numbers need context before they become evidence.

A correct number attached to the wrong outcome, time point or reference group is still wrong. MetaWiz keeps those semantic checks visible.

✓

Check the article location, row, column, group, unit and time point before accepting a candidate value.

⇄

Route values only when the statistical structure and clinical outcome are compatible with the destination dataset.

2×

Keep reviewer drafts separate and create the canonical analysis record only after reconciliation.

Supported evidence patterns

Continuous outcomesSample size, mean and SD structures for MD or SMD workflows.

Dichotomous outcomesEvents and totals for RR or OR analyses.

Time-to-event effectsHazard ratios with confidence intervals and model context.

2×2

Diagnostic accuracyTP, FP, FN and TN structures for diagnostic evidence synthesis.

Reviewer → final datasetControlled

Independent extraction

Independent extraction

Resolve discrepancies

Analysis-ready row

Start with citation screening and keep evidence, reviewer decisions, datasets and analysis connected from beginning to end.