There is an urgent, unmet need for biomarkers of risk for Alzheimer’s disease related dementia (ADRD) suitable for routine examination in naïve patients. Studies suggest MRI-derived brain connectome may contain salient information about brain health status. However, it has yet to be tested in sufficiently large samples whether connectome can be used to predict reliably diagnosis of ADRD. Here we performed high-throughput computational analysis using structure and diffusion MRI in a clinical cohort (N=211) to estimate morphometry and connectome. Our results show potential utility of data-driven machine learning models using large-scale MRI-derived brain phenotypes in classifying ADRD, particularly structural connectome.