Katharina Buczek
Throughout countless industries, artificial intelligence (AI) has shown that it can revolutionize business practices– saving time and resources while mitigating mistakes. Healthcare is no different.
AI has already performed successful medical diagnostic tests. For instance, categorizing images based on pathological changes with high accuracy.
Still, it had been more challenging to teach AI to examine patient conditions– which are time-varying– and make treatment suggestions.
But in a recent study conducted by researchers at TU Wien and the Medical University of Vienna, this very feat was accomplished.
Using extensive data collected from the intensive care units (ICUs) of various hospitals, the team developed artificial intelligence that was able to provide treatment suggestions for sepsis patients who required intensive care. Remarkably, analyses also revealed that artificial intelligence performed better than humans in terms of decision quality.
“In an intensive care unit, a lot of different data is collected around the clock. The patients are constantly monitored medically. We wanted to investigate whether these data could be used even better than before,” said Professor Clemes Heitziner of the Institute for Analysis and Scientific Computing at TU Wien.
Healthcare professionals already make their treatment decisions based on various parameters– which must be taken into account before any plans are suggested and followed. Although, artificial intelligence is able to process more parameters than a human could, which might even result in better decision-making.
So, in the team’s study, they used “reinforcement learning,” which is a type of machine learning.
“This is not just about simple categorization– for example, separating a large number of images into those that show a tumor and those that do not– but about a temporally changing progression, about the development that a certain patient is likely to go through,” Heitzinger explained.
Sign up for Chip Chick’s newsletter and get stories like this delivered to your inbox.
