How Should Artificial Intelligence Support Clinician Decision Making?

Human brain in microchip.
For decades, clinicians and researchers have been interested in seeing what role computers can play in complex clinical decision making.

A recent viewpoint article published in JAMA focused on decision support systems designed for interactive use by clinicians to aid them in making decisions in complex clinical situations. As technologies evolve, decision support tools are becoming more widespread across areas like medical devices. However, clinical decision support systems (CDSSs) have been challenged in their credibility, even with decades of research that provides evidence to the contrary, and these systems have not seen widespread adoption.

Edward H. Shortliffe, MD, PhD, of Columbia University, and Martin J. Sepulveda, MD, ScD, of IBM Research, shared their views on reasons why CDSSs have not been implemented. One major concern is lack of usability. This includes the efficiency and intuitiveness of a CDSS, which should enable it to blend into a busy clinical environment. Along with this, use of a CDSS should complement, not replace, a clinician. A CDSS should also, they say, be relevant to the specific domain and reflect understanding of specific questions clinicians are likely to have and offer a strong scientific foundation.

Drs Shortliffe and Sepulveda stressed that even with the aid of a CDSS, there is no “right answer,” and that a clinician and a CDSS can reach different conclusions and either or neither could be correct. There are no formal regulatory standards for interactive decision support software leading to physician action, although analytic decision software implemented as part of a medical device does have regulatory standards and therefore a monitoring system during use is essential to identify potential issues.

Related Articles

There is ongoing work, they concluded, to recognize the full capabilities required of an effective CDSS. However, they note that these systems offer value if they continue to be improved and effectively integrated into clinical settings.


Shortliffe EH, Sepúlveda MJ. Clinical decision support in the era of artificial intelligence. JAMA. 2018;320(21):2199-2200.

This article originally appeared on Medical Bag