Artificial Intelligence in Healthcare
The growing demand for more efficient and effective healthcare workflows has led to an increasing number of artificial intelligence (AI) products in the sector. This trend is especially pronounced in the field of Radiology, where AI solutions are being developed to assist with image analysis, diagnosis, and workflow management. However, with over 200 AI products in the field of Radiology currently available, selecting the most useful and appropriate algorithm has become a complex and challenging task for healthcare providers.
Purpose of the Project
Project AIR aims to simplify this selection process by providing a structured framework for evaluating different AI algorithms. By testing these algorithms across multiple centers using a diverse, multi-center dataset, Project AIR offers a standardized comparison that helps healthcare institutions make more informed decisions when choosing AI solutions. The goal is to streamline the adoption of AI in clinical practice by providing transparent, evidence-based evaluations.
Framework
Vendors: Project AIR invites vendors who have developed commercially available AI algorithms for healthcare to participate in the evaluations. These algorithms must have a valid Conformité Européenne (CE) mark, ensuring they meet European safety, health, and environmental protection standards.
Grand Challenge Platform: To facilitate this project, Amazon Web Services (AWS) will be used to host the algorithms within Docker containers, ensuring a secure and scalable environment for testing.
Reader Study: A gold standard evaluation will be conducted using a reader study approach. In this process, radiologists will assess the performance of each algorithm, with the results being validated against a reference standard, such as a majority vote or expert consensus.
The general Project AIR protocol as well as the use cases specific protocols can be found here.
Do you want to contribute as a vendor?
Vendors who are interested in participating in Project AIR, either by submitting their existing AI algorithms for evaluation or by contributing to new or updated use cases, are encouraged to get involved. At present, we are working on various use cases, and new ones can be initiated or updated as part of the project’s continuous development.
For more information or to initiate a use case, please contact us.