AI and Bronchoscopy

Photo by Franck V. on Unsplash

This is the first of several posts about the role of artificial intelligence and the future of interventional pulmonology*.  I am confident our field will change immensely in the years ahead, and that artificial intelligence will not only change how we learn and perform procedures but also how we interact with patients. The sooner we embrace these changes, and build partnerships with industry as well as colleagues from other disciplines such as computer engineering, ethics, psychology, philosophy, physics, mathematics, and business administration, the easier it will be to integrate new developments into clinical practice.

Artificial intelligence has many definitions. A quick Google search provides “the theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.” Wikipedia expands on this definition, adding that AI “describes machines (or computers) that mimic “cognitive” functions that humans associate with the human mind, such as “learning” and “problem-solving.”

This requires us to familiarize ourselves with the phraseology computer engineers use to describe the learning process, but which is not necessarily foreign to many educators.

From a developmental perspective, AI uses symbolic, connectionist, and other models of learning that are, in fact, similar to how the human brain works. Just as there are several types of knowledge, AI does not rely on only one developmental approach to provide results. This is elegantly explained in a 1990 article by Marvin Minsky (AI magazine, summer 1991), in which he explains how the sentence “ Mary gave Jack the book” prompts the human brain to produce a visual representation of the act, a tactile representation of the experience, a script-sequence of what it means ‘to give’, and various assumptions about Jack, Mary, and the book. Similarly, artificial intelligence must employ not one but several different strategies to provide a result.

Some results are methodology—based on algorithmic and probabilistic approaches. Computer-based interpretation of pulmonary function tests, image-pattern recognition for accurate computed tomography scan interpretation, and patient management protocols based on decision-tree and data-driven statistical algorithms are simple examples of how artificial intelligence brings complex knowledge instantaneously to our fingertips. No longer required to memorize facts and figures, or integrate history/clinical exam/laboratory findings into patterns learned through a prolonged patient-care apprenticeship, doctors will change their practice habits accordingly.

  • Please subscribe to Colt’s Corner to automatically receive email notification of future posts. Sign up with your name and email on the NEWSLETTER button on the Bronchology International home page at  www.bronchoscopy.org.