Developing Algorithm-Based Recommendations in the ACR: Defining a New Process

Area of Science:

• Radiology
• Medical Informatics
• Clinical Guidelines

Background:

• The American College of Radiology (ACR) previously used informal consensus for algorithm-based recommendations, limiting scalability and formal guideline designation.
• The lack of a formal, evidence-driven process hindered incidental findings and other algorithms from meeting the National Academy of Medicine's Trustworthy Guidelines criteria.
• Limited referrer and policymaker adoption was observed due to the absence of a standardized, evidence-based development process.

Purpose of the Study:

• To introduce a new, formal, evidence-driven process for developing algorithm-based imaging recommendations within the ACR.
• To overcome the limitations of the previous informal consensus method, including scalability and generalizability.
• To enable ACR algorithms to meet criteria for designation as clinical guidelines.

Main Methods:

• Formation of a multi-stakeholder committee with ACR approval, including representatives from key ACR committees, commissions, and diverse practice settings.
• Development of a formal, evidence-driven methodology for creating algorithm-based imaging recommendations.
• Ensuring the process is generalizable for adoption by various ACR committees.

Main Results:

• A new, formal, evidence-driven process for algorithm-based imaging recommendations has been successfully developed and documented.
• The new process addresses the limitations of informal consensus, enhancing scalability and generalizability.
• This formalized approach is designed to facilitate compliance with the National Academy of Medicine's Trustworthy Guidelines criteria.

Conclusions:

• The newly developed evidence-driven process provides a robust framework for creating ACR algorithm-based imaging recommendations.
• This formalized process is expected to improve the quality, adoption, and guideline designation of ACR recommendations.
• The methodology is generalizable and can be adopted by other ACR committees for their evidence-based recommendation development.