2023 Alzheimer's Association Research Grant to Promote Diversity (AARG-D)
AI Assisted Neuropathological Assessment of Sporadic Alzheimer's Cases
How can machine learning help understand Alzheimer’s?
Diego Sepulveda-Falla, Ph.D.
University Medical Center Hamburg-Eppendorf
Hamburg, Germany
Background
Alzheimer’s is associated with specific brain changes including the accumulation of the proteins beta-amyloid and tau into abnormal plaques and tangles, respectively. Some people develop Alzheimer’s due to specific changes or variations in their DNA that they inherited, which is known as “familial” Alzheimer’s. Others develop Alzheimer’s without a clear genetic link, known as “sporadic” Alzheimer’s. Though individuals with familial Alzheimer’s typically develop Alzheimer’s at a younger age, there is still a lot of variability in age of onset, and there is still much to learn about the changes in the body and brain that promote the disease.
In initial studies, Dr. Diego Sepulveda-Falla and colleagues applied a machine learning algorithm (a type of artificial intelligence) to analyze features of brain samples, such as beta-amyloid plaques, from 126 individuals who had familial Alzheimer’s. The analyses of brain changes combined with data about the individual’s health history led the team to identify four Alzheimer’s subtypes in this group.
Research Plan
Dr. Diego Sepulveda-Falla and colleagues will now apply what they learned from the individuals with familial Alzheimer’s and analyze data from 457 individuals with sporadic Alzheimer’s cases with their machine learning algorithm. They aim to optimize the algorithm and to determine which variables are most important to identify clusters of patients with similar health histories and brain changes.
Impact
Results from this study will contribute to an understanding of the different ways Alzheimer’s can progress, which may help diagnose and predict the course of the disease in the future.