Our investigation begins with an effort to harness GSMM to systematically describe the metabolic state in AD on a global, network level. We do this by employing a method termed integrative Metabolic Analysis Tool, which incorporates gene expression into a GSMM to predict metabolic flux activity. This method has already been shown to successfully predict tissue specific metabolic activity in several healthy human tissues, including the brain. iMAT incorporates gene expression to predict global metabolic flux activity that is the most consistent with known constraints across the entire metabolic network, and reflects post transcriptional modifications that are not evident in the raw expression data. We utilized a relatively large dataset of gene expression microarrays from the cortex of AD patients and elderly controls, which we integrated with the human metabolic model to study the metabolic changes in AD. This model-based genome-scale view of AD metabolism leads to the identification of various pathways whose activities are altered significantly in AD, and importantly, are not revealed by standard pathway enrichment analysis of the raw gene expression Berberrubine solely, in a model-free manner. We next predict novel biomarkers for AD by comparing predicted uptake and secretion fluxes of various metabolites as the disease progresses. Finally, we predict perturbations in the metabolic network that can transform the metabolic state of AD back closer to a healthy state, highlighting new taraxasteryl-acetate potential metabolic drug targets for AD that may work on a global, network level. As mentioned earlier, differences between gene expression levels and enzyme flux activities as predicted by iMAT can indicate whether enzyme activity is post-transcriptionally increased or decreased compared to the original mRNA levels. To test the metabolic descriptions we have obtained, we compared the predicted alterations in enzyme activities to the measured protein levels of these enzymes, according to proteomic data from temporal cortex of AD patients. Reassuringly, we find significant overlap between predicted and experimentally determined differences in the levels of these proteins.