The identification of coherent biomarkers, essential for clinical and therapeutic progress, requires an understanding of either the relevant pathogenic processes or, at the very least, of the parameters that need to be monitored. But, syndrome-dominated conceptual thinking has become a barrier to understanding the biological processes linked to diseases characterized by clinical and therapeutic heterogeneity. As a result, current biomarkers of cognitive disorders are much too numerous, too heterogenous, and too variable to serve useful purposes. This leads to an untenable situation that precludes coherent therapeutic developments since it effectively prevents defining what could constitute valid biological, clinical and therapeutic approaches. How to escape from this situation? The problem could be partly resolved by adopting the much wider views allowed by “system-wide” approaches: in effects, by constructing predictive theoretical models of what could constitute pathological cognitive processes. This, naturally, requires the integration of massive amounts of highly heterogenous and often conflicting information. This chapter aims to provide a necessarily brief overview of the concepts, the breadth of data and the variety of network dynamics that will have to be considered while proposing a functional, experimentally validated model-building approach that could be fruitfully utilized.