We will develop new analysis approaches based on state-of-the-art machine-learning/AI models that allow Alzheimer’s Disease researchers to test for the absence of a genetic association signal. The approaches will be particularly robust against the effects of model selection uncertainty and potential model misspecifications. We will provide a software implementation of our approach in the family-based association tests (FBAT) program and apply our approach to the data of the Alzheimer’s Genome Project. We will publish all of our analysis results.
