Skip to content

AIRBDS AI-Readiness Dataset Scoring Metric

A practical guide to evaluating bioscience datasets for AI/ML use

Developed by the AIRBDS Working Group, AIBIO-UK · Funded by BBSRC · Licensed CC BY 4.0


Overview

Questions this tutorial answers:

  • What does the AIRBDS metric measure, and why does it matter?
  • Should I use the CSV (spreadsheet) or YAML (text file) format?
  • How do I work through all 28 questions and calculate a grade?
  • How do I submit my completed review?

Learning Objectives

By the end of this tutorial, you will be able to:

  1. Describe what the AIRBDS metric assesses and how it is scored
  2. Choose the format (CSV or YAML) that suits your technical background
  3. Complete a full dataset review and assign a grade (Caution / Bronze / Silver / Gold)
  4. Submit a completed review to the repository

Who this is for: Researchers, data curators, and repository managers working with bioscience datasets

Skill levels covered: Beginner (CSV path) · Intermediate (YAML path)

Estimated time: 30–60 minutes per dataset review


What is the AIRBDS Metric?

The AIRBDS metric is a structured checklist of 28 Yes/No questions that assesses whether a bioscience dataset is ready for use in AI and machine learning workflows. Questions cover four areas:

Scope Questions What it checks
Infrastructure ACM-1 – ACM-10 Access, licensing, unique identifiers, version control
Metadata ACM-11 – ACM-17 Bias documentation, standards, preprocessing, provenance
Content ACM-18 – ACM-23 Completeness, consistency, format
Ethics ACM-24 – ACM-28 Consent, privacy, security, data protection

Each answer contributes to a weighted score. Datasets receive one of four grades:

Grade Meaning
🔴 Caution Fails one or more Critical criteria — serious limitations for AI/ML use
🟤 Bronze Passes most Critical questions (≥ 7/8)
Silver Passes all Critical + ≥ 50% of Important questions
🟡 Gold Passes all Critical and Important + ≥ 50% of Optional questions

How to use this tutorial

Start here → Chapter 1: Getting Started — Choose Your Format

Chapter 1 explains the difference between the CSV and YAML formats and helps you choose the right one based on your experience level. You will then follow either:

  • Chapter 2 — CSV walkthrough (beginner, no coding required, Excel or Google Sheets)
  • Chapter 3 — YAML walkthrough (intermediate, text editor and command line)

Citation

If you use this metric in your research, please cite:

AIRBDS Working Group, AIBIO-UK. (2025). AIRBDS AI-Readiness Dataset Scoring Metric (v0.3). GitHub. https://github.com/AIBIO-UK/airbds-metric

Full citation metadata is in CITATION.cff.