Article type
Abstract
Background:
Guideline development is a resource-intensive process that requires substantial time and effort. Collaboration among guideline development groups has the potential to reduce duplication of efforts, optimize resource utilization, and enhance the quality and efficiency of guideline development processes. Here, we illustrate a successful application of a collaborative approach to guideline development between the BMJ Rapid Recommendations Group and the American Gastroenterological Association (AGA) on the use of an artificial intelligence (AI) tool called Computer Aided Detection (CADe) for colonoscopy.
Method:
As one of the most actively researched applications of AI, CADe has been rapidly gaining attention in the field of gastroenterology as a potential tool to improve colorectal cancer-related outcomes. Both the BMJ Recommendation group and the AGA prioritized this topic for guideline development. Acknowledging that both groups adhere to rigorous methodology with an emphasis on transparency, broad panel representation, and strict management of conflict of interest helped to establish an informal collaboration.
Results:
The evidence to inform both guidelines included a published systematic review/meta-analysis and a modeling study from the European-funded OperA (Optimising Colorectal Cancer Prevention Through Personalised Treatment with Artificial Intelligence) project. The co-development process allowed for sharing of evidence profiles with judgments for the certainty of the evidence and collaboration with members of both teams to analyze evidence to inform the evidence-to-decision framework, namely patient values and preferences. An important consideration was the acknowledgment that different assumptions may be needed for the modeling study and that differences in health care systems, feasibility, cost/resource implications, accessibility, and equity factors may necessitate adaptation for the non-European context.
Conclusion:
With rapidly changing evidence and the need for trustworthy recommendations tailored to local contexts, collaboration and resource pooling among guideline groups offer a promising path forward. Several factors helped promote successful collaboration: (1) use of the GRADE framework and adherence to international standards, (2) collective approach to evidence synthesis with contributions from members of both panel groups to create efficiencies, and (3) co-representation across the two guideline groups to facilitate alignment. This example demonstrates how successful collaboration between groups reduces redundancies in effort and enhances guideline development.
Guideline development is a resource-intensive process that requires substantial time and effort. Collaboration among guideline development groups has the potential to reduce duplication of efforts, optimize resource utilization, and enhance the quality and efficiency of guideline development processes. Here, we illustrate a successful application of a collaborative approach to guideline development between the BMJ Rapid Recommendations Group and the American Gastroenterological Association (AGA) on the use of an artificial intelligence (AI) tool called Computer Aided Detection (CADe) for colonoscopy.
Method:
As one of the most actively researched applications of AI, CADe has been rapidly gaining attention in the field of gastroenterology as a potential tool to improve colorectal cancer-related outcomes. Both the BMJ Recommendation group and the AGA prioritized this topic for guideline development. Acknowledging that both groups adhere to rigorous methodology with an emphasis on transparency, broad panel representation, and strict management of conflict of interest helped to establish an informal collaboration.
Results:
The evidence to inform both guidelines included a published systematic review/meta-analysis and a modeling study from the European-funded OperA (Optimising Colorectal Cancer Prevention Through Personalised Treatment with Artificial Intelligence) project. The co-development process allowed for sharing of evidence profiles with judgments for the certainty of the evidence and collaboration with members of both teams to analyze evidence to inform the evidence-to-decision framework, namely patient values and preferences. An important consideration was the acknowledgment that different assumptions may be needed for the modeling study and that differences in health care systems, feasibility, cost/resource implications, accessibility, and equity factors may necessitate adaptation for the non-European context.
Conclusion:
With rapidly changing evidence and the need for trustworthy recommendations tailored to local contexts, collaboration and resource pooling among guideline groups offer a promising path forward. Several factors helped promote successful collaboration: (1) use of the GRADE framework and adherence to international standards, (2) collective approach to evidence synthesis with contributions from members of both panel groups to create efficiencies, and (3) co-representation across the two guideline groups to facilitate alignment. This example demonstrates how successful collaboration between groups reduces redundancies in effort and enhances guideline development.