Alzheimer's disease (AD) is characterized by peri-neuronal amyloid plaque and intra-neuronal neurofibrillary tangles. These aggregates are identified by the immunodetection of "seed" proteins (A&#
x3b2;<sub>1-42</sub> and hyperphosphorylated tau, respectively), but include many other proteins incorporated nonrandomly. Using click-chemistry intra-aggregate crosslinking, we previously modeled amyloid "contactomes" in SY5Y-APP<sub>Sw</sub> neuroblastoma cells, revealing that aspirin impedes aggregate growth and complexity. By an analogous strategy, we now construct amyloid-specific aggregate interactomes of AD and age-matched-control&#
xa0;hippocampi. Comparing these interactomes reveals AD-specific interactions, from which neural-network (NN) analyses predict proteins with the highest impact on pathogenic aggregate formation and/or stability. RNAi knockdowns of implicated proteins, in <i>C.&#
xa0;elegans</i> and human-cell-culture models of AD, validated those predictions. Gene-Ontology meta-analysis of AD-enriched influential proteins highlighted the involvement of mitochondrial and cytoplasmic compartments in AD-specific aggregation. This approach derives dynamic consensus models of aggregate growth and architecture, implicating highly influential proteins as new targets to disrupt amyloid accrual in the AD brain.