Advances of high-order interactions in the human brain : applications in aging and neurodegeneration

The human brain generates a large repertoire of spatio-temporal patterns, which support a wide variety of motor, cognitive, and behavioral functions. The most accepted hypothesis in modern neuroscience is that each of these representations is encoded in different brain networks. From MRI, networks can be defined anatomically (“structural connectivity”-SC) or functionally (“functional connectivity”-FC). Interestingly, while SC is by definition pairwise (white matter fibers project from one region to another), FC is not. In this thesis we have focused on the study of high-order interactions (HOI) that occur in functional networks, beyond the existing statistical relationships in pairs of regions. When evaluating the interacting n-plets, from triplets to order n, a novel type of statistical interdependencies appear, namely the synergistic and redundant interactions, which are inaccessible when evaluating interacting pairs. The study of these HOI in the human brain in aging and neurodegeneration is the purpose of this thesis. Starting from the O-Information formalism, we have systematically analyzed synergistic and re- dundant interactions in functional networks of the human brain. In the first part of this thesis, we have applied this formalism to the aging brain and have found a higher preva- lence of redundant interdependences in older participants compared to younger ones, and this effect occurs in all orders of interaction within regions located in the prefrontal and motor areas, thus involving working memory, motor and executive functions. In the second part of the thesis, we have built a neurobiological-realistic computational model of the whole-brain that incorporates SC and FC data. Our model shows that, related to aging, variations in functional patterns can be explained by changes in SC, which neurodegenerate as we age. Based on this finding, we propose a simple nonlinear neurodegeneration model that is representative of healthy (non-pathological) aging and that reproduces the age variations that occur in the HOI structure of functional data. Finally, in the last part of this thesis, we have applied our formalism to a clinical population, and in particular to a cohort of older patients with Frontotemporal Dementia (FTD), comparing the redundant and synergistic patterns that occur in the HOI of these brains, in comparison with a control group of healthy population, well matched in age, sex and years of education with the FTD cohort. For this chapter, we have developed a new statistical tool that allows us to detect clusters that are significantly different between groups and where the interaction is predominantly synergistic or redundant. For our particular case of FTD, redundant triplets were found in higher-order vision networks, default mode, and salience network. Similarly, synergistic triplets were found in the primary auditory cortex. Together, the results obtained in this thesis pave an avenue of multiple possibilities in the study of HOI as informational markers in the high-order functionality of the brain, and how its alterations could reveal new organizational aspects of the human brain in health and disease.