Does it calculate effect sizes automatically?

Yes, it can derive Hedges' g and standard errors automatically from primary statistics such as means, standard deviations, and sample sizes.

What is the Universal Meta-Analysis Codebook?

It is a structured framework for extracting, standardizing, and verifying statistical data from research papers using an AI-human collaborative workflow.

How does it handle AI extraction errors?

The skill uses a confidence-based triage system (High, Medium, Low) and requires mandatory human verification for all extracted rows to ensure 100% accuracy.

Can I extend the codebook for specific research fields?

Yes, it supports context-specific extensions that allow researchers to define custom moderator variables for domains like Clinical Trials or Educational Technology.

Which agents does this skill integrate with?

It is designed to integrate with Diverga C5, C6, and C7 agents for automated extraction, data integrity guarding, and final validation gates.

Universal Meta-Analysis Codebook

Name: Universal Meta-Analysis Codebook
Author: HosungYou

byHosungYou

0•

数据科学与机器学习

Standardizes meta-analysis data extraction through a multi-layered AI-human collaboration framework and automated statistical provenance.

The Universal Meta-Analysis Codebook is a specialized tool designed to streamline the complex process of data extraction for systematic reviews and meta-analyses. It utilizes a robust four-layer architecture—covering metadata, core statistics, AI provenance, and human verification—to ensure high-integrity datasets. By integrating with RAG and OCR technologies via specialized agents, it automatically extracts critical values like Hedges' g and sample sizes from PDFs, applies confidence scoring to every data point, and manages a rigorous human-in-the-loop verification queue, making it an essential utility for researchers requiring production-grade statistical analysis.

主要功能

01Four-layer data structure for identifiers, statistics, AI provenance, and human verification.

02Automated extraction from research PDFs using RAG and OCR with confidence tracking.

03Context-specific extensions for specialized domains like Clinical Trials and EdTech.

04Integrated triage system that prioritizes human review based on AI extraction confidence.

05Automated calculation of Hedges' g and standard errors from raw statistical inputs.

060 GitHub stars

使用场景

01Standardizing data extraction workflows across large-scale systematic review teams.

02Automating the identification of effect sizes and moderator variables in academic literature.

03Ensuring 100% data integrity in meta-analyses through structured verification pipelines.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add hosungyou/diverga universal-ma-codebook

For use in Claude.ai and ChatGPT

Download Skill