An artificial intelligence system makes neurological diagnoses in record time

Researchers in the US have successfully tested a new artificial intelligence system to test brain imaging and diagnose acute brain conditions such as stroke, bleeding, and hydrocephalus.

The system takes just 1.2 seconds -far less than doctors- to get it diagnosed. The scientists and physicians at the Medical School and the Mount Sinai Hospital in New York, headed by Eric Oerman of the Neurosurgery Department, made the relevant publication in the medical journal “Nature Medicine”.

“With a diagnosis time of just 1.2 seconds, the system can alert doctors in time for a critical finding, which would otherwise remain in wait for minutes or even hours. Our goal is to develop medical artificial intelligence that will solve clinical problems and improve patient care”,

said Oerman. It is the first study that uses artificial intelligence to detect a wide range of acute neurological events. Researchers used 37,236 brain tomographs to train a neural network of deep learning to diagnose. The system was then tested in a blind, randomized and controlled clinical trial with regard to the correct diagnosis speed as compared to experienced radiologists. It was found that the average diagnosis time from the computational algorithm was on average 150 times less than doctors. AISINAI team researchers, who are developing the system, will be ready for clinical use in about two years.   Source: www.cnn.gr