Skip to Main Content

We have a new app!

Take the Access library with you wherever you go—easy access to books, videos, images, podcasts, personalized features, and more.

Download the Access App here: iOS and Android


First introduced in 1956 at a Dartmouth College conference, the concept of artificial intelligence (AI) is not new. However, with the recent advances in computing power and data storage, artificial intelligence has been integrated into our daily lives. From autocomplete to Amazon recommendations, AI algorithms have influenced how we search, shop, and get around. With the transition to electronic medical records, whole-genome sequences, and high-resolution images, medicine has also entered the era of big data and has the potential benefit from the promise of AI. In medicine, AI promises not only increased convenience but also provides comprehensive and personalized care. Although hoping to achieve no human error, the accuracy depends on data entry as a source of potential error.

The earliest work of AI in medicine occurred in the 1970s. The first NIH-sponsored AI in Medicine workshop occurred in 1975 at Rutgers University.1 In 1976, the first AI prototype in medicine, the CASNET model, was presented at the Academy of Ophthalmology meeting. This model can process patient and disease-specific inputs and produce treatment recommendations for those with glaucoma.2 In the late 2000s, the growth of AI is fueled by significant advances in the field of computer science, such as increased processing speed and power. With the development of the convolutional neural network (CNN), a type of deep learning network, the application of AI in high-resolution medical images became possible. For example, after training on 14,884 3D optical coherence tomography scans, AI applications can make referral recommendations for a range of sight-threatening retinal diseases better than the experts.3 In nephrology, 700,000 medical records were used to create an AI algorithm that can predict the development of acute kidney injury 48 hours before traditional clinical care.4

Although the application of machine learning (ML) to clinical medicine is in its infancy, many algorithms have been developed to predict clinical outcomes such as sepsis, dementia, readmission, and mortality postchemotherapy.5–8 Dr Topol, a cardiologist and an expert in translational research, has predicated that AI technology will be incorporated into every clinical practice.9 However, despite its seemly unlimited potential, the current clinical application of AI is still a work in progress.


By definition, AI describes any design system that demonstrates the properties of human intelligence, such as reasoning, learning, adaptation, or sensory understanding. In its most basic form, an AI algorithm is a set of algorithms that produces output based on input data without human interference between both these points. The human role is during the input stage and in creating the algorithms that help analyze the data for the needed output. There are many subfields under the AI umbrella, including ML, deep learning, and CNN (Figure 57-1). The field of AI can also be subdivided into a physical branch which consists of physical assistive robots for care, surgery, or drug ...

Pop-up div Successfully Displayed

This div only appears when the trigger link is hovered over. Otherwise it is hidden from view.