Researchers use AI to detect Alzheimer's in advance

However, as with any disease, treatment is much easier with a diagnosis made in advance so that even the remedies become more effective. Now researchers at the University of California at San Francisco have taught artificial intelligence to recognize the early symptoms of Alzheimer's in the brain six years in advance, much faster than a common diagnosis.

The diagnosis of Alzheimer's is made through an analysis of glucose levels in different areas of the brain. The detail is that accuracy is difficult to achieve, especially since the changes in substance level are subtle.

Thus, doctors involved in the project sought the help of an artificial intelligence algorithm to detect the discrete changes caused by Alzheimer's. At the end of the experiment, the result was satisfactory: according to Dr. Jae Ho Sohn, the application of deep learning is very efficient in identifying subtle and diffuse processes. The algorithm was trained with approximately 2,000 images and then was tested again with images from other exams. In the first attempt, AI reached the 92% accuracy mark; already in the second, that number rose to 98%. SOURCE (S) BGR