RADHawk-an AI-based knowledge recommender to support precision education, improve reporting productivity, and reduce...

Area of Science:

• Artificial Intelligence in Medical Education
• Radiology Training and Development
• Precision Education Systems

Background:

• Developed RADHawk (RH), the first AI-derived knowledge recommender for radiology.
• RH provides personalized, real-time educational content for radiology residents and fellows.
• Aims to augment knowledge and establish precision education in radiology.

Purpose of the Study:

• Assess the impact of an AI-based knowledge recommender (RH) versus traditional methods.
• Evaluate effects on radiology trainee reporting time, quality, and cognitive load.
• Compare learning experiences and attitudes toward AI in radiology education.

Main Methods:

• Mixed methods prospective study comparing RH users (intervention) to non-users (control).
• Assessed technology acceptance, report quality, time, cognitive load, and learning attitudes via questionnaires and simulated reporting.
• Used non-parametric regression and Mann-Whitney tests for outcome comparisons (P<0.05).

Main Results:

• Intervention group showed significant reduction in reporting time (-162s/case) and increased accuracy (+14%).
• Lower mental demand, effort, and frustration reported in the intervention group.
• >78% of intervention group rated RH positively; 89% requested continued access.

Conclusions:

• RH, the first AI knowledge recommender in radiology, reduces reporting time and improves accuracy.
• RH decreases trainee workload and mental demand, enhancing the learning experience.
• High acceptance suggests RH's potential for self-directed learning and precision radiology education.