Could machine learning help predict an individual’s risk of developing Alzheimer’s?
Joseph Giorgio, Ph.D.
University of California
Berkeley, CA - United States
Researchers use many strategies to measure an individual’s brain function and to track brain changes associated with cognitive diseases. One strategy used in Alzheimer’s research is to measure levels of dementia-related proteins, such as tau and beta-amyloid (two hallmark proteins found in individuals with Alzheimer’s). Tau and beta-amyloid can be measured in biological samples, such as blood, as well as via a specialized brain scans called 3-dimensional positron imaging tomography (PET).Another brains can approach called structural magnetic resonance imaging (sMRI)can measure overall brain function include monitoring brain shape and volume.
Each of these techniques has the potential to identify early brain changes associated with Alzheimer’s. However, it can be difficult to integrate findings across different approaches and connect them to clinical symptoms, which is an essential step to understanding an individual’s overall risk of developing Alzheimer’s.
Dr. Joseph Giorgio and colleagues will develop an advanced computer science technique called machine learning to predict brain changes and cognitive decline associated with Alzheimer’s. They will evaluate how well biological markers (biomarkers) measured in blood, brain scan results, and clinical data can be combined to predict hallmark brain changes associated with Alzheimer’s. Dr. Giorgio will first combine cognitive testing and brain scan data to understand the level of Alzheimer’s related brain changes in more than 5,000 individuals. Dr. Giorgio will then use machine learning to identify brain scan features that correspond to each stage of Alzheimer’s. Next, the researchers will use similar approaches to connect cognitive testing data with sMRI brain scan features. Finally, Dr. Giorgio’s team will test whether the brain scan features they identified can be used to predict cognitive decline.
This study could be used to combine complex brain scan data with clinical findings to help understand early Alzheimer’s related brain changes and to predict future brain changes in individuals at risk of developing Alzheimer’s.
Back to Top