Having already explored the impact of brain similarity networks in cognitively normal individuals across the aging spectrum, we now aim to extend this research to individuals affected by Alzheimer’s disease (AD). The specific aims of the project are: 1. Investigate whether brain similarity networks derived from T1-weighted structural MRI can detect early brain changes in preclinical AD, potentially serving as sensitive biomarkers before clinical symptoms emerge. 2. Evaluate the predictive power of brain similarity networks in comparison to conventional biomarkers, such as cortical thickness, volumetric measures, and fluid biomarkers, in forecasting cognitive decline. 3, Assess the ability of brain similarity networks to predict conversion from cognitively normal status to mild cognitive impairment (MCI) and AD, and determine whether combining brain similarity with fluid biomarkers enhances predictive accuracy. 4. Explore the structural disruptions captured by brain similarity, with a particular focus on cortical layer II of the cytoarchitectonic lamina, to better understand the microstructural basis of neurodegeneration. 5. Integrate brain similarity with multimodal imaging and clinical data to develop robust, biologically meaningful biomarkers for early detection, monitoring, and stratification of AD progression. As the operations involved in developing graph neural networks in Tensorflow are computationally intensive, we need a robust server to effectively support our analysis. Additionally, since we are working with medical images that take up more space than other types of data used in conventional machine learning algorithms, having an additional storage resource would be highly beneficial. Finally, the data xthat would be used is properly anonymized.