[Correspondence] AI-supported mammography screening: measuring benefit

The MASAI trial represents an important step in the field of mammography screening. Jessie Gommers and colleagues report that an artificial intelligence (AI)-supported workflow achieved non-inferiority in the interval cancer rate. This means that AI-supported screening was not worse than the standard approach in detecting cancer that appears between two screening examinations. Furthermore, the study noted higher sensitivity, indicating AI's improved ability to correctly identify the presence of disease. Specificity, meaning the ability to correctly determine the absence of disease, remained unchanged. Despite these positive findings, several methodological issues warrant careful consideration before widespread adoption of this method. These issues require attention to ensure the reliability and safety of AI-supported screening in practice.