Risk Assessment in Population-Based Breast Cancer Screening
Mikael Eriksson, Emily F. Conant, Despina Kontos, Per Hall
Abstract
Yala et al 1 reported on the multinational validation of the Mirai artificial intelligence-based mammographybased breast cancer risk model.The study included 62,185 screening mammograms and 3,815 incident cancers diagnosed within 5 years of study entry.Concordance indices ranged from 0.75-0.84for the 5-year risk model across seven screening sites.It was concluded that the risk tool could offer broad and equitable improvements in breast care.The authors advance the important inclusion of imagebased data to further refine existing models for the assessment of breast cancer risk.The identification of women who, after a negative screen, may benefit from supplemental screening could lead to earlier detection of breast cancers and potentially improved prognoses as previously described. 2The goal of equitable improvements in care is also essential for broad implementation.We, therefore, read this study with great interest and applaud the authors for their undertaking.However, there are several aspects of this study that require clarification.