The AI Path to Deeper and More Accurate Medicine –

The subtype of AI, deep learning, has recently been used for a variety of medical applications, particularly image interpretation, across many disciplines. It is early, but these deep neural networks have considerable transformative potential to promote accuracy, streamline workflow, synthesize data for both clinicians and patients, promote patient autonomy, and greatly enhance the patient–clinician relationship. To date, prospective validation studies are quite limited, and we need to confront such vital issues of how AI can unintendedly worsen inequities and discrimination. The goal, which may be considered counter-intuitive, is to use this technology to make medicine more human.

View the full lecture by Eric Topol on the VideoCast by NIH.