skills.rest
mims-harvard
mims-harvardOFFICIAL
Explore all skills by this author →
Website𝕏@marinkazitnikGitHub

tooluniverse-meta-analysis

Pool effect sizes across multiple studies with fixed- or random-effects meta-analysis and heterogeneity statistics.

1.7k|254|Updated Mar 3, 2025
Search Tags:#systematic review#meta-analysis#evidence synthesis#heterogeneity#effect size#forest plot#random-effects
One-click install
npx skills add https://github.com/mims-harvard/ToolUniverse --skill tooluniverse-meta-analysis
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: tooluniverse-meta-analysis
Source: https://github.com/mims-harvard/ToolUniverse/tree/main/plugins/tooluniverse/skills/tooluniverse-meta-analysis
Command: npx skills add https://github.com/mims-harvard/ToolUniverse --skill tooluniverse-meta-analysis

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Combining results from multiple studies into one pooled estimate is error-prone: ratio measures must be log-transformed, standard errors must be derived from confidence intervals, and heterogeneity must be quantified before trusting the pooled number. This Skill handles the full evidence-synthesis workflow so systematic reviews and multi-cohort analyses produce a defensible pooled estimate.

Core Features & Use Cases

  • Effect-size conversion: Converts OR/RR/HR with CIs, regression betas, two-group means (Hedges' g), proportions, and correlations (Fisher z) into the (effect, SE) pairs the pooling step needs.
  • Fixed- and random-effects pooling: Runs DerSimonian-Laird random-effects or fixed-effects models via MetaAnalysis_run or the bundled Python script, returning pooled estimates, CIs, p-values, and per-study weights.
  • Heterogeneity assessment and forest plots: Reports Q, I-squared, and tau-squared with interpretation guidance, and prints a text forest plot with per-study weights and the pooled diamond.
  • Use Case: You have odds ratios with 95% CIs from three cohorts studying the same exposure. Feed them to the helper script as a CSV and get a pooled OR back-transformed to the ratio scale, plus I-squared and a forest plot for your systematic review.

Quick Start

Ask the AI to pool the effect sizes from your extracted studies into a single random-effects meta-analysis estimate with heterogeneity statistics and a forest plot.

Frequently Asked Questions about tooluniverse-meta-analysis

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I pool odds ratios from multiple studies in a meta-analysis?▼

Log-transform each odds ratio and derive its standard error from the 95% CI using (ln(upper) - ln(lower)) / 3.92, then pool with a random-effects model. The helper script accepts a CSV with or, ci_low, ci_high columns and returns the pooled OR already back-transformed.

Fixed-effects vs random-effects meta-analysis: which should I use?▼

Use fixed-effects only when all studies estimate the same true effect, such as exact replications. Use random-effects (DerSimonian-Laird) when studies differ in population or design, which is the usual real-world case; it is the conservative default and widens the CI to reflect between-study variance.

What does I-squared mean in meta-analysis heterogeneity?▼

I-squared is the percentage of variation across studies due to true heterogeneity rather than chance. Values of 0-25% indicate low heterogeneity, 25-50% moderate, 50-75% substantial, and above 75% considerable, where pooling may be inappropriate and study differences should be explained instead.

Why does MetaAnalysis_run give wrong results for odds ratios?▼

MetaAnalysis_run pools whatever scale it receives, so feeding raw odds ratios silently produces a wrong pooled value. You must pass log-transformed effect sizes and standard errors, then exponentiate the pooled result and CI bounds back to the ratio scale; the helper script handles this automatically.

Can I assess publication bias with only a few studies?▼

No. Funnel plots and Egger's test require at least about 10 studies to be reliable. With fewer studies, small-study effects cannot be assessed, so report this limitation explicitly rather than interpreting an underpowered test.

Related Skills

Explore More Education & Research Skills→
astoreyaiastoreyai

synthesis-matrix

Organize extracted study data into structured evidence matrices for systematic reviews.

Community
Intermediate
nahisahonahisaho
3

scientific-meta-analysis

Aggregate study effect sizes into pooled estimates with heterogeneity and bias metrics.

Community
Advanced
Estrella-231Estrella-231

scientific-critical-thinking

Audit study design, bias risks, and statistical validity of scientific claims.

Community
Advanced
choxoschoxos
6

Meta-Analysis Methods in R

Automates meta-analyses in R using the metafor package.

Community
Advanced