How Vetra works

Our guiding principles

The researcher in the middle

RAISE

Responsible use of AI in evidence SynthEsis

From question to protocol

Design

Shape your research question, define the review context and iterate until you have a protocol that meets your goals.

  1. Framework
  2. Research question
  3. Objectives and selection criteria
  4. Time horizon, languages and study types

From question to protocol

Design

Shape your research question, define the review context and iterate until you have a protocol that meets your goals.

  1. Framework
  2. Research question
  3. Objectives and selection criteria
  4. Time horizon, languages and study types

What the guidelines say

Leading methodological guidelines recommend defining a clear research question, eligibility criteria, and review methods in advance. Recording these decisions in a protocol helps maintain methodological consistency and justify subsequent amendments.

Your role

You define the scope of the review, examine Vetra’s suggestions, and validate the research question, objectives, and eligibility criteria. You decide when the protocol accurately reflects what you intend to investigate.

Responsible use under RAISE

AI use during study design should be justified and transparent. Generated suggestions and subsequent changes should be identifiable so that the final protocol reflects deliberate, reviewed methodological decisions.

Which papers are relevant to my review?

Screening

Find potentially relevant studies and review each rationale to decide what to include; you can provide the full texts.

  1. Title and abstract screening, followed by full-text screening
  2. Automatic retrieval of Open Access papers
  3. Document decisions for PRISMA reporting

Which papers are relevant to my review?

Screening

Find potentially relevant studies and review each rationale to decide what to include; you can provide the full texts.

  1. Title and abstract screening, followed by full-text screening
  2. Automatic retrieval of Open Access papers
  3. Document decisions for PRISMA reporting

What the guidelines say

Leading guidelines recommend applying predefined eligibility criteria, using independent assessment procedures where appropriate, and resolving discrepancies with documented reasons. Selection decisions should be recorded to make the process traceable.

Your role

You take part in a complementary, independent AI–researcher assessment: Vetra conducts an assessment and you can make your own without being influenced by its recommendation. You then compare the outcomes, resolve discrepancies, and confirm which studies to include.

Responsible use under RAISE

AI can support study selection provided its performance is evaluated and appropriate oversight is maintained. Automated decisions, discrepancies, and human interventions should be identifiable and documented so the reliability of screening can be assessed.

What data lies behind the papers?

Data extraction

Extract structured information from each included paper for later synthesis.

  1. Configure the extraction template
  2. Extract structured information

What data lies behind the papers?

Data extraction

Extract structured information from each included paper for later synthesis.

  1. Configure the extraction template
  2. Extract structured information

What the guidelines say

Leading guidelines recommend structured extraction forms tailored to review objectives and verification procedures that reduce errors. Data should be collected consistently and remain linked to their original sources.

Your role

You configure which information to extract and check the results suggested by Vetra. You compare data against the original articles, correct possible errors, and validate the information before using it in later stages.

Responsible use under RAISE

AI-extracted information should be verifiable. Preserve data provenance and distinguish between information directly identified in studies, AI-generated interpretations, and human corrections to avoid introducing unverified results.

Evidence quality

Study appraisal

Interpret findings in light of how each study was designed and conducted to assess the strength of the evidence.

  1. Set appraisal scales by study type
  2. Determine the quality of the evidence gathered
  3. Use internationally recognized scales, such as JBI and Cochrane

Evidence quality

Study appraisal

Interpret findings in light of how each study was designed and conducted to assess the strength of the evidence.

  1. Set appraisal scales by study type
  2. Determine the quality of the evidence gathered
  3. Use internationally recognized scales, such as JBI and Cochrane

What the guidelines say

Leading guidelines recommend assessing methodological limitations using tools appropriate to each study design. Judgments should follow explicit criteria and include reasons that allow every assessment to be understood and reviewed.

Your role

You select the assessment tools, examine Vetra’s suggested rationales, and decide whether they accurately reflect each study. You can adjust judgments and resolve disagreements before confirming the final assessment.

Responsible use under RAISE

AI-assisted methodological judgments require particular attention to reliability. They should be grounded in verifiable information, acknowledge limitations, and retain reviewable justifications rather than automatically delegating evidence interpretation to the system.

From findings to conclusions

Synthesis

Brings together the collected evidence to provide a clear, well-grounded view of the findings.

From findings to conclusions

Synthesis

Brings together the collected evidence to provide a clear, well-grounded view of the findings.

What the guidelines say

Leading guidelines recommend choosing synthesis methods based on study characteristics and the comparability of findings. Before pooling data, differences in populations, interventions, measures, and designs, as well as limitations of the available evidence, must be considered.

Your role

You determine which results can be grouped and which analyses appropriately address the research question. You review Vetra’s tables and results, interpret the findings, and validate the conclusions.

Responsible use under RAISE

AI should help interpret evidence without distorting its meaning. Synthesized findings must be traceable to the original data, distinguishing observed results from AI-assisted interpretations and avoiding conclusions that go beyond the evidence.

From conclusions to manuscript

Writing

Bring the findings together in a reviewable manuscript you can edit and complete with supporting materials.

  1. Generate a draft from documented work
  2. Produce supplementary materials

From conclusions to manuscript

Writing

Bring the findings together in a reviewable manuscript you can edit and complete with supporting materials.

  1. Generate a draft from documented work
  2. Produce supplementary materials

What the guidelines say

Leading reporting guidelines recommend clearly communicating a systematic review’s methods, findings, and limitations. The manuscript should faithfully represent the work performed, remain consistent across sections, and make the research assessable and reproducible.

Your role

You review the manuscript prepared with Vetra’s assistance, verify findings and references, and refine the scientific interpretation. You choose what to include and approve the final version, remaining responsible for its conclusions.

Responsible use under RAISE

AI involvement in drafting the manuscript should be disclosed transparently. Identify the tools used, the tasks they supported, and how their outputs were verified, while respecting scientific integrity, authorship, and human accountability.