KAIST develops world’s first guidelines on use of AI in healthcare

Artificial intelligence experts in Korea have put forward technology guidelines to ensure the credibility of AI in the healthcare sector for the first time in the world.

According to the Korea Policy Center for the Fourth Industrial Revolution at the Korea Advanced Institute of Science and Technology Sunday, it has developed “Using Artificial Intelligence to Support Healthcare Decisions: A Guide for Society” in collaboration with the Institute for the Public Understanding of Risk at the National University of Singapore and U.K.-based Sense About Science.

The guide contains examples of AI use in the medical and healthcare sector, including disease forecast through analysis of medical images and big data, and reduction of time required for new drug development. The guide also emphasizes that if data from learning is partially omitted or excluded, AI can be biased, and that if it is used for purpose different from the original purpose, judgement can be ill-advised.

The guide also contains five criteria for ensuring the credibility of AI. First, the source of data should be clear; data should be used in accordance with the purpose; limitations and hypotheses should be clearly defined; bias of data should be clarified; and whether proper tests were conducted in a real environment or not should be checked.

“We expect that the guide will play an important role for people to recognize the limitations of AI and elements that require improvement,” said Kim So-young, head of the Korea Policy Center for the Fourth Industrial Revolution.


shinjsh@donga.com