Breast cancer screening stands at a technological crossroads as healthcare systems worldwide grapple with radiologist shortages and increasing demand for mammography services. The promise of artificial intelligence to augment or even replace human interpretation has captured significant attention, yet the real-world evidence remains fragmented across disparate studies with varying methodologies and populations.
This comprehensive systematic review analyzed the growing body of research on AI applications in mammography screening, examining both standalone AI systems and AI-assisted radiologist workflows. The analysis encompassed diverse study designs from randomized controlled trials to real-world implementation cohorts, focusing on critical performance metrics including sensitivity, specificity, recall rates, and cancer detection capabilities. The review methodology was thorough, searching major medical databases and clinical trial registries from 2012 through 2025 without language restrictions.
While the complete findings await publication, this systematic approach represents a crucial step toward understanding AI's true potential in population-based breast cancer screening. The evidence synthesis arrives at a pivotal moment when healthcare systems are making substantial investments in AI technology without comprehensive guidance on optimal implementation strategies. Current AI systems show varying performance across different populations and screening contexts, with some demonstrating comparable or superior detection rates to experienced radiologists while others struggle with false positive rates that could overwhelm healthcare resources. The review's inclusion of workflow outcomes alongside pure diagnostic accuracy metrics reflects growing recognition that successful AI integration requires consideration of practical implementation challenges, radiologist acceptance, and system-wide efficiency gains rather than simply matching human performance on isolated test cases.