README.md

metadata

title: Cardio-AI Best Practice Checklist
emoji: 🫀
colorFrom: red
colorTo: pink
sdk: gradio
sdk_version: 4.44.1
app_file: app.py
pinned: false

Cardio-AI — Best Practice Checklist (Gradio Space)

An interactive checklist inspired by the paper:

Evaluating artificial intelligence-enabled medical tests in cardiology: Best practice.
IJC Heart & Vasculature (2025). DOI: 10.1016/j.ijcha.2025.101783
ScienceDirect: https://www.sciencedirect.com/science/article/pii/S2352906725001861

Use this Space to plan, review, and report supervised ML studies in cardiology (with examples from cardiac electrophysiology).

Features

• Cohort/splits, outcomes/ground truth, classification, regression, reference models, explainability, validation, and governance sections.
• TRIPOD-AI mapping helper.
• Reviewer Mode: one-click reviewer-style strengths/concerns/verdict summary.
• Export: Markdown (built-in) and PDF (via reportlab); optional filename override.
• Save/load JSON state.
• Preset for a typical ECG AF classifier internal validation package.
• “Exemplar Figures & Prompts” tab outlines what to report for PRC/ROC/BA/Confusion Matrix.

How to use

1. Tick items across tabs to match your study.
2. Save → JSON to keep progress for later editing (you can Load it back).
3. Export as Markdown/PDF to generate a report; a download tile appears right next to the button, and a collapsible preview shows the Markdown content.
4. Use the TRIPOD-AI tab to map items to reporting sections.
5. Use Reviewer Mode to generate a referee-style summary for internal review.

Requirements

• gradio>=4.44.1
• reportlab (optional; needed for PDF export)

Notes

• The checklist is a practical companion to the above paper; tailor items to your dataset, devices, journal policies, and regulator expectations.
• This tool does not replace formal statistical review or regulatory guidance; it helps structure reporting and internal QA.