The problem it solves
After extraction, the researcher needs to decide which results are comparable, which outcomes are synthesised, which studies are included in each analysis, and how to communicate certainty without losing the trail of earlier decisions.
What Vetra does
- Groups results by outcome, population, intervention, comparator and effect measure.
- Helps prepare tables and figures from previously verified data.
- Flags inconsistencies between studies, units, follow-up times or outcome definitions.
- Supports documentation of GRADE judgments when needed, without replacing the researcher's assessment.
What the researcher decides
The researcher decides which results are clinically comparable, which analyses are appropriate and how to interpret the certainty of the evidence. The AI can organise and flag, but it should not generate final conclusions without critical review.
What is recorded
- Outcomes included in each synthesis.
- Studies and data used in each table or figure.
- Justification for exclusions, groupings or transformations.
- GRADE judgments, changes made and the responsible reviewer.
GRADE and meta-analysis
RAISE reminds us that AI use in synthesis and conclusions must be transparent and defensible. Vetra improves this by breaking GRADE into an interactive table that lets the researcher review each domain, check the evidence and correct the judgment before closing the final synthesis.
The researcher can review each part, modify it and justify it before closing the synthesis. Meta-analysis will be performed when the study data allow it and the comparability between results is sufficient.