Although we lack a full understanding of the particular cascade of pathological events that give rise to Alzheimer's disease (AD), the convergence of multiple threads in neuroscience is cause for optimism. In particular, we see the emergence of i) in vivo PET tracers that illustrate the distribution of protein misfolding, such as tau and beta-amyloid, which complements more conventional imaging, genetics and measurements behavioral; ii) advanced multimodal analytical strategies for identifying disease subtypes and temporal stages as well as modeling disease progression; and iii) a culture of open sharing of neuroscientific data which, with the support of advanced neuroinformatics platforms, freely brings a wealth of AD-related data to the global community. This conference will discuss our recent progress in these areas which translates into a move away from the clinical categorization of disease in favor of a more mechanistic approach to its categorization. This strategy should promote better-informed design of biomarkers and therapeutic interventions, better patient stratification and, ultimately, personalized medicine.