<2-3 sentences summarising what the project addresses and solves, the target audience, and the benefit.> This project provides a comprehensive, AI-driven skill package for medical imaging researchers, addressing the entire research lifecycle from ideation to grant applications. It targets researchers aiming for high-impact publications in journals like Radiology and the Nature series, offering a structured workflow to enhance study design, methodology, statistical rigor, and manuscript preparation, thereby mitigating common reviewer critiques and increasing adoption success.

How It Works

The radiology-skills package functions as a modular Codex skill set, comprising 22 specialized "virtual consultants." It operationalizes best practices derived from real-world high-impact publication and funding experiences, focusing on Radiology and Nature series journal standards. The system guides users through a full research pipeline, emphasizing adherence to reporting guidelines (e.g., CLAIM, TRIPOD+AI), robust statistical validation, and proactive identification of methodological flaws, such as data leakage and insufficient external validation. Its novelty lies in translating expert reviewer perspectives directly into actionable audit steps.

Quick Start & Requirements

• Primary install / run command: Clone the repository (git clone https://github.com/huang-sir1/radiology-skills.git) and copy the radiology-skills/ directory into ~/.codex/skills/. Requires the Codex environment.
• Non-default prerequisites and dependencies: Codex environment.
• Links: Project Repository

Highlighted Details

• Full Research Lifecycle Coverage: Spans ideation, data annotation, modeling, statistical analysis, writing, submission, response to reviewers, and grant applications.
• Dual Journal Specification: Integrates specific formatting and reporting requirements for both Radiology family and Nature series journals.
• Proactive Reviewer Simulation: Audits research design, methodology, and reporting against common critical reviewer questions before submission.
• Expert-Driven Rules: Developed by authors with first-author publications in Radiology, ensuring practical, experience-based guidance.
• Integrity Commitment: Explicitly avoids fabricating data, statistics, or qualifications, marking uncertainties clearly.

Maintenance & Community

• Contributors: Core development by authors with first-author publications in Radiology.
• Updates: Recent updates (July 2026) include comprehensive README localization, expanded international grant support (NIH R01, ERC, Wellcome), and enhanced Nature series journal compliance.
• Community: No specific community links (e.g., Discord, Slack) are provided in the README.

Licensing & Compatibility

• License type: The specific open-source license is not detailed in the README.
• Compatibility notes: No explicit notes regarding commercial use or compatibility with closed-source projects are present. The tool is geared towards academic research workflows.

Limitations & Caveats

Many skills are in "Beta" or "Draft" status, indicating ongoing refinement and potential for edge cases. It functions as an assistive tool, not a replacement for domain experts like biostatisticians or clinical collaborators; users must independently verify current guidelines and eligibility. The tool explicitly disclaims providing medical advice, and users are solely responsible for the accuracy and completeness of all research materials.