Systematic Reviews, Automated

Finish your systematic
review in hours, not months

Revv runs the full systematic review pipeline — screening, extraction, and meta-analysis — and delivers evidence tables in hours. Every decision the AI makes is yours to trace and review.

Submit your research question

$250 for a typical review · No account required · Results delivered to your inbox

Used by researchers at

Western UniversityUniversity of TorontoSunnybrook

A typical systematic review takes 6 to 18 months and over 1,000 hours of manual work. Teams read thousands of abstracts one by one, copy numbers into spreadsheets by hand, and redo the process when new studies publish. Most of the labor is mechanical. The judgment calls — what to include, how to rate bias, whether the evidence holds — those are the parts that matter.

Screening

Thousands of abstracts,
reviewed in minutes

The AI reads every abstract against your criteria and suggests include or exclude with its reasoning. You make the final call.

Abstract Screening · 24 of 127 studies reviewed

Comparison of sedation protocols for pediatric imaging

Williams, Park, & Hassan · 2022 · Anesthesia & Analgesia

Background: Pediatric MRI frequently requires sedation to minimize motion artifacts. Comparative data on sedation protocols remain limited. Methods: Multicenter RCT comparing propofol (n=84), sevoflurane (n=81), and dexmedetomidine (n=79) in children aged 1 to 7 years. Primary outcomes were image quality and time to discharge. Results: Image quality adequate in 96% propofol, 91% sevoflurane, 94% dexmedetomidine. Emergence agitation: 2.4% vs 14.8% vs 1.3% (P<0.001).

AI Suggestion

Include

Multicenter RCT comparing sedation agents in pediatric patients undergoing MRI. Matches population (children 1 to 7), intervention (propofol, sevoflurane, dexmedetomidine), and study design (RCT) criteria. Reports primary outcomes of interest (image quality, discharge time, adverse events).

Data Extraction

Every number traceable
to its source

Click any extracted value to see exactly where it came from in the original paper. Page, paragraph, and surrounding context.

Data Extraction · Williams et al. 2022

Arm	n	Image Quality (%)	Discharge (min)	Agitation (%)	Risk of Bias
Propofol	84	96.4	38 (28–52)	2.4	Low
Sevoflurane	81	91.4	51 (38–67)	14.8	Some concerns
Dexmedetomidine	79	94.3	62 (45–78)	1.3	Low

Source

Median time to discharge was shortest for the propofol group at 38 minutes (IQR 28 to 52), compared with 51 minutes for sevoflurane and 62 minutes for dexmedetomidine (Table 3).

Williams et al. 2022, p. 847, Results, paragraph 3

Field Details

VariableDischarge time

Value38 (28–52)

UnitMinutes

StatisticMedian (IQR)

ConfidenceHigh

Meta-Analysis

From extracted data to
publication-ready outputs

Forest plots, summary of findings tables, and GRADE assessments generated automatically from your extracted data.

Meta-Analysis · Discharge Time (Propofol vs Sevoflurane)

Study

Mean Difference (95% CI)

MD [95% CI]

Williams 2022

-13.0 [-19.2, -6.8]

Zhang 2023

-11.5 [-16.8, -6.2]

Koroglu 2021

-15.3 [-23.1, -7.5]

Chen 2023

-9.8 [-15.4, -4.2]

Patel 2024

-14.1 [-21.6, -6.6]

Pooled estimate

-12.6 [-16.1, -9.1]

← Favours propofolFavours sevoflurane →

I² = 18% (low heterogeneity)

τ² = 2.4

P < 0.001

5 studies, n = 487

Outcome	Studies	Participants	Effect (95% CI)	Certainty
Time to discharge	5	487	MD -12.6 min [-16.1, -9.1]	High
Image quality (adequate)	4	402	RR 1.05 [0.99, 1.11]	Moderate
Emergence agitation	5	487	RR 0.16 [0.06, 0.43]	High
Serious adverse events	3	310	RD 0.00 [-0.02, 0.02]	Low

How It Works

Your review, step by step

Describe your protocol. Revv runs the pipeline. You review and approve each stage before moving on.

Define your protocol

Paste your research question, inclusion criteria, and outcomes. Revv structures it into a searchable protocol.

Search and deduplicate

Queries run across databases. Results are merged on DOI, then fuzzy-matched on title and year.

Screen abstracts and full texts

Two-pass screening: abstracts first for obvious disqualifiers, then full texts for definitive inclusion.

Extract data and assess risk of bias

Variables generated from your protocol. Each study’s data extracted per arm, comparison, or study level.

Meta-analysis and GRADE

Pool effects, generate forest plots, assess heterogeneity, and rate certainty of evidence per outcome.

Export

PRISMA flow diagrams, forest plots, funnel plots, summary of findings, and risk of bias tables.

Why Revv

Why researchers use Revv
instead of ChatGPT

General-purpose LLMs

Fabricate effect sizes, confidence intervals, and sample sizes. Confidently.

Provide no way to trace a number back to the source. Not for you, not for your reviewers, not for the journal.

Can describe PRISMA, risk of bias, and GRADE fluently. Apply none of them. The output looks methodical without being methodical.

Work one prompt at a time. A 500-study review needs a pipeline, not a longer chat window.

Revv

Links every number to the page, paragraph, and table in the source PDF.

Runs screening, extraction, and synthesis as separate stages. You approve each one before the next starts.

Follows the protocols your field requires. RoB 2, ROBINS-I, QUADAS-2, and GRADE are built into every step, not bolted on after.

Runs the same way whether you have 15 studies or 1,500.

Finish your systematicreview in hours, not months