AfCAI Association for Clinical Artificial Intelligence

DECIDE-AI Checklist

Interactive reporting checklist for the early-stage live clinical evaluation of decision support systems driven by artificial intelligence.

Generate Reports
Save progress
Checklist controls
Generic items only
0%
Items rated
0
Reported
0
Not Reported
0
N/A
0
Unrated
AI-specific items only
0%
Items rated
0
Reported
0
Not Reported
0
N/A
0
Unrated
Combined
0%
Items rated
0
Reported
0
Not Reported
0
N/A
0
Unrated
About This Tool: The DECIDE-AI Checklist

What is it for?
This tool is a web-based checklist implementing DECIDE-AI, the reporting guideline for the Developmental and Exploratory Clinical Investigations of DEcision support systems driven by Artificial Intelligence. DECIDE-AI is a stand-alone, stage-specific guideline for reporting the early, small-scale, live clinical evaluation of AI-based decision support systems, focusing on proof of clinical utility at small scale, safety, human factors, and preparation for larger summative trials. It comprises 17 AI-specific reporting items (28 sub-items) and 10 generic reporting items. By using this checklist, reviewers can systematically evaluate whether an early-stage clinical AI study transparently reports these elements.

This web tool is provided for educational and review purposes only and does not constitute formal editorial, regulatory, methodological, clinical, or legal advice. Please consult the original publications for the full official guidance.

Sources and attribution
Checklist items are derived from the DECIDE-AI checklist (Table 2). Explanatory summaries have been independently rewritten and condensed from the DECIDE-AI Explanation & Elaboration for educational and review purposes.

1. Vasey, B., Nagendran, M., Campbell, B. et al. Reporting guideline for the early-stage clinical evaluation of decision support systems driven by artificial intelligence: DECIDE-AI. BMJ 377, e070904 (2022). doi:10.1136/bmj-2022-070904. Supplementary appendix 1: Explanation and Elaboration.
This checklist is not endorsed by the original authors.