With an eye to AI and autonomous diagnosis
In Digital medicine you'll find the article on an artificial intelligence-based diagnostic of diabetic retinopathy. It's the first AI system approved by FDA last April, and represents the begining of new aids for medical decision making. Therefore, there is a room for supporting decision making with AI, but maybe one day autonomous diagnosis could be the issue, who knows.
The prespecified sensitivity end point agreed with the FDA was 85.0% and this was met with a point estimate of primary sensitivity of 87.2%. However, the confidence intervals of this estimate were 81.8–91.2% (that is, spanned the superiority end point). The study also employed an intention-to-screen protocol; however, 40 participants successfully enrolled in the study were excluded from analysis as their images were subsequently found to be insufficient quality to be graded by the image reading center. The authors attempt to address this by considering a worst-case scenario where all such images are incorrectly graded and repeating the analysis. In this approach the sensitivity would
be 80.7% (76.7–84.2%).
Although deep learning will not be a panacea, it has huge potential in many clinical areas where high dimensional data is mapped to a simple classification and for which datasets are potentially stable over extended periods. As such, it will be incumbent on healthcare professionals to become more familiar with this and other AI technologies in the coming years to ensure that they are used appropriately.
Alice Francis - Coco Baca Bum Bum
