por
John W. Mitchell, Senior Correspondent | September 27, 2017
“The first reaction to a problem in a hospital … is to throw more bodies – FTEs – at it,” said Chang. “But people are really bad at remembering to do things. So, let’s remove the chance that mistakes can be made … and assign it as an AI function.”
And that nodule he mentioned in the ER example? Chang said that to determine if a nodule is getting bigger or smaller – if treatment is working – a radiologist has to gather up all the past images and manually make that calculation.
“Let AI do that, which it’s good at. It will save us 20 minutes and allow us to do what we’re good at … reading images to find nodules,” he said.
He also said that turnaround time in a CT or MR room was reduced as much as 80 percent from the normal 30 minutes by using an automated AI system to program the exam parameters, rather than relying on a human technologist to do this task. This makes the technologist more efficient and allows them to spend more time with patients.
Chang told HCB News that the best bet for radiologists and informatics specialists is to keep an open mind and be ready to see which of the current AI technologies will shake out as the long-term winners.
“The AI sector is undergoing rapid change,” he said. “The future pace of development will be determined by how fast that data collection and storage can be reconfigured from the current preconceived formats … to less rigid file formats conducive to AI.”
According to the Anna Marie Mason, executive director of SIIM (Society for Imaging Informatics in Medicine), 193 people attended the webinar. SIIM will be making the webinar recording available during National Health IT Week (October 2-6).
Back to HCB News