Facilitating the use of routine data to evaluate artificial intelligence solutions: lessons from the NIHR/RCR data...

Area of Science:

• Medical Imaging and Artificial Intelligence
• Health Informatics
• Clinical AI Evaluation

Background:

• Artificial intelligence (AI) is rapidly advancing in radiology, necessitating collaboration for ethical and responsible development.
• Ensuring AI's progress in clinical practice requires addressing challenges in its evaluation and deployment.
• Three workshops were organized by the NIHR/RCR committee to foster knowledge sharing among stakeholders.

Purpose of the Study:

• To describe the outcomes of the first workshop focused on using routine data for AI solution evaluation in radiology.
• To identify key challenges and strategies for the ethical and effective use of AI in medical imaging.
• To promote collaboration between academics, industry, and clinical leaders.

Main Methods:

• The study reports on the findings from a workshop involving key stakeholders in AI and radiology.
• Discussions focused on practical aspects of using routine clinical data for AI evaluation.
• Key topics included ethical considerations, data curation, de-identification, and data sharing.

Main Results:

• Essential ethical considerations for AI development and deployment were identified.
• Methods for data curation, storage, and de-identification were detailed.
• Secure data-sharing strategies, quality assurance, data access committees, and patient perspectives were explored.

Conclusions:

• Facilitating the use of routine data is crucial for evaluating AI solutions in radiology.
• Addressing ethical concerns, data management, and secure sharing are vital for responsible AI advancement.
• Collaboration and consideration of patient perspectives are key to enhancing AI in healthcare.