The shortcomings of the area under the receiver operating characteristic curve (ROC AUC) have been well understood for years. When used to assess Medical Imaging AI, it’s primary weakness is the lack of focus on clinical impact. Why then, have Medical Imaging AI companies have been slow to move beyond AUC? In this new article, our CEO describes how moving beyond ROC AUC, and emphasizing “cut points” can improve our ability to gauge the clinical impact of Medical Imaging AI algorithms.