Artificial intelligence, MRI combination achieves 94% accuracy in predicting dementia

A form of artificial intelligence combined with MRI scans of the brain predicted with 94% accuracy whether individuals with a specific type of early memory loss would go on to develop Alzheimer's diseases or other forms of dementia, according to a new study. University of Florida researchers studied 55 participants diagnosed with amnestic mild cognitive impairment -- a condition that is a known precursor to Alzheimer's disease and marked by signs like forgetting conversations or misplacing items. By applying a support vector machine model algorithm to a 45-minute MRI brain scan of brain structures and brain activity, researchers reported they were able to predict progression from amnestic MCI to dementia with over 94% accuracy. The algorithm delivered 92.7% accuracy when applied to a 10-minute brain scan. Fourteen participants developed dementia and 41 remained stable over a 15-month study period.