Summing It All Up

Abstract

Much of the AI that exists today in clinical medicine is based on the application of machine learning methods to Big Data, such as imaging, outcomes predictions, and decision support. We focus on the key information that clinicians should understand when presented with AI-based clinical systems, including reviewing key definitions and a framework for evaluation of AI systems in clinical medicine. We then discuss where and how these systems can be expected to evolve, from integrating AI-based point solutions to real-time learning within medical practice environments. Beyond this, we expect broad adoption that allows learning and integration across systems, subject to new changes in healthcare data privacy regulations and technologies that preserve privacy across systems. Finally, we discuss the use of AI in medical situations without big or structured data. If done well, AI promises to improve the clinical experience for patients and providers by tightly integrating treatment, quality improvement, and research in a continuous learning system. However, the rapid pace of new technology creates risks that clinicians must understand as they begin to accept AI systems into their clinical practice.

Ähnliche Arbeiten