AI-Assisted Chest X-Ray Interpretation in Resource-Limited Settings: LuAna Stepped-Wedge Trial Protocol

Area of Science:

• Radiology and Medical Imaging
• Artificial Intelligence in Healthcare
• Clinical Decision Support Systems

Background:

• Chest radiography is a widely used, cost-effective imaging tool limited by anatomical complexity and expertise variability.
• Artificial intelligence (AI) offers potential to enhance diagnostic accuracy, identify subtle findings, and reduce errors in chest X-ray interpretation.
• AI integration can address challenges in resource-constrained settings, but requires robust real-world validation.

Purpose of the Study:

• To evaluate if AI assistance improves general practitioners' detection of radiographic findings on chest X-rays in adults with respiratory complaints.
• To compare the diagnostic accuracy of AI-assisted interpretation versus unaided interpretation, using expert radiologist consensus as the reference.

Main Methods:

• A multicenter, stepped-wedge, cluster-randomized clinical trial following CONSORT-AI and SPIRIT-AI guidelines.
• The intervention uses Lung Analysis (LuAna), an AI-powered chest X-ray interpretation tool, across 9 Brazilian cities.
• General practitioners' interpretations will transition from unaided to AI-assisted monthly, benchmarked against thoracic radiologists.

Main Results:

• The study is ongoing, with data collection from January 2026 to September 2026.
• Data analysis is projected to conclude by November 2026, with results expected by January 2027.
• Preliminary center preparation indicates readiness for enrollment.

Conclusions:

• AI-assisted chest radiography is expected to enhance clinical decision-making and diagnostic pathways for respiratory conditions.
• The findings will provide real-world evidence on the clinical utility of AI in routine imaging workflows.
• Successful implementation could offer a scalable model for AI integration, improving access to specialist care globally.