PROJECT SUMMARY Alzheimer's disease (AD) is a devastating neurodegenerative disease that deeply impacts the quality of life both socially and financially for affected ones and their relatives. Despite extensive clinical and genomic studies, the exact mechanisms of development and progression of AD remain elusive. Microglia and other myeloid origin cells, collectively known as human brain immune cells (HBICs), have been identified to play crucial roles in the pathogenesis of AD. This is supported through genetic association studies, where many of the common and rare risk loci affect genes that are preferentially or selectively expressed in HBICs, emphasizing the pivotal role of the innate immune system in AD. In the parent grant 1R01AG065582, we utilize fluorescence-activated cell sorting to isolate CD45+/CD11b+ HBICs from human brain fresh tissue. We then apply innovative neurogenomics and single-cell approaches to generate comprehensive, high-throughput, multi-omics molecular profiles of HBICs from 300 donors at different stages of AD. These remarkable resources can provide critical insights into the role of immune cells in AD by increasing our mechanistic understanding of dysfunction in AD risk loci. One critical component that is currently not addressed in the parent grant is to apply innovative genomic approaches using AI/ML techniques, which can harmonize the signals from different omics modalities and offer a novel insight into the role of microglia and other immune cells in AD. In this Supplement, to increase the utility of the data, we propose to develop and maintain a shared resource of high-dimensional HBIC omics data for AI/ML applications. In addition, we propose to build a multi-scale integrative deep learning model leveraging single-cell omics data, to demonstrate the utility of the resource and serve as a benchmark for others to provide a quantitative measure of performance. This model will help us to identify protective and neuroinflammatory HBIC subpopulations and colocalize transcriptomic and regulatory signatures at different stages of AD. The proposed work will address potential challenges in the development of AI/ML applications. We propose: (1) identifying and removing potential sources of technical variations and normalize the data (2) uniformly processing and preparing fully annotated AI/ML-ready resource in a self- contained form for rapid prototyping with modern AI/ML tools (3) sharing and collaborating ideas using an open forum using AD knowledge base portal. Successful completion of the proposed studies will: (1) facilitate access to large-scale, multidimensional datasets on HBICs for AI/ML applications; (2) accelerate researches for an increased mechanistic understanding of the onset and progression of AD; (3) provide systems-level insights about transcriptional regulation in HBICs and AD pathogenesis using integrative AI/ML model; (4) provide a prioritized list of significant loci and genes for future mechanistic stu...