While artificial intelligence is being widely used in a multitude of fields from accounting to farming, one of the biggest applications of AI is in health care, where one of the biggest problems is a lack of review.
The capacity of artificial intelligence to diagnose or predict disease risk is developing rapidly. This week in Nature, the topic of AI is prevalent, as AI models that scan retinal images for diseases have recently been unveiled. The biggest problem, however, lies within the studies of the AI’s performance, or lack thereof.
In terms of performance studies, the standard procedure is to perform numerous tests that can be replicated in a progression of studies that culminates in large scale trials, much like clinical testing. However, many related to the field feel that these tests on AI are simply not being performed to enough of a degree, as some claims have still yet to be peer reviewed. While this may feel minor, these small details are crucial to further development.
Though it is exciting to develop technologies such as this, it is vastly important to thoroughly study its performance, and describe the results in great detail. This approach may take longer, but it could prevent faulty tech, and even save lives, especially in the case of health care. The extra time taken to carefully review the performance of these technologies could, quite literally, save lives of many people, while also saving the time it would take to fix the problems that would arise with faulty equipment.
Pax Per Scientiam 