Mount Sinai AI uncovers new brain analysis method to predict dementia, Alzheimer's disease

Current methods for diagnosing Alzheimer’s disease rely on a complex combination of self- and caregiver-reported symptoms, a physical examination and either a PET scan or a spinal tap to look for evidence of amyloid plaque build-ups in the brain.

But a new artificial intelligence-based method may make the diagnostic process a much more objective one. Developed by researchers from New York City’s Mount Sinai and described in a study published this week in the journal Acta Neuropathologica Communications, the AI was trained to predict the presence of cognitive impairment with only minimal human supervision.

To develop the AI, the researchers began by analyzing the cellular and anatomical makeup of the brain’s medial temporal lobe and frontal cortex. From there, they fed a deep learning algorithm slide images of brain autopsy tissues collected from more than 700 elderly patients and taught it to look for changes in the amount of myelin acting as a protective layer around the brain’s nerves.

With that training, an AI model was able to pinpoint certain characteristics of the myelin that were linked to cognitive impairment. In addition to decreasing amounts of the protective layer, that included seeing the myelin scattered in a non-uniform pattern across the brain tissue, with the diminishing amounts focused mainly in the white matter, the area of the brain linked to learning and neurological function.

Overall, the study’s authors wrote, the trained AI was able to predict the presence of cognitive impairment “with a modest accuracy that was significantly greater than chance.”

The deep learning model may not yet be ready for widespread use—especially since follow-up analyses weren’t entirely able to understand how the AI reached some of its predictions—but the researchers expressed optimism that it and other similar algorithms could one day be used to improve diagnosis of Alzheimer’s disease and other neurodegenerative conditions.

Future studies may see the researchers dig into the reasoning behind the AI’s conclusions to boost their accuracy and reliability, and could also involve scaling up the model to uncover additional pathways to improve screening for cognitive impairment.

“Leveraging AI allows us to look at exponentially more disease-relevant features, a powerful approach when applied to a complex system like the human brain,” said the study’s co-corresponding author Kurt Farrell, Ph.D., an assistant professor at Mount Sinai’s Icahn School of Medicine. “It is critical to perform further interpretability research in the areas of neuropathology and artificial intelligence so that advances in deep learning can be translated to improve diagnostic and treatment approaches for Alzheimer’s disease and related disorders in a safe and effective manner.”