favorite3Specifically, we find that happiness and positive attitude have the most significant jump when using social network structure features in addition to health behavior and demographic data.
favorite3As in the case of stress, the table shows that using social network structures can improve prediction performance for these 3 health and wellness variables.
favorite4Specifically, we examine the relationship between social network topological properties including degree, number of triangles, clustering coefficient, closeness centrality and betweenness centrality for each node (individual) in the participant network and in the whole network and health behavioral variables including heart rates, steps and activity states.
favorite5For that reason, we selected the survey taken in Fall 2016, which contains questions about wellness states - stress, happiness, positive attitude and self-assessed health - and covers most of our participants (380 subjects) as this is the first semester in which all three tiers are present in the study.
favorite3This physical data is obtained from the Fitbit metrics which include health-related behavioral variables such as heart rate, step and activity states.
favorite10Then, we formalize the notion of context-specific dependence (CSD) and deterministic context-specific dependence (DCSD) for hierarchical GNMs. We discuss how to improve the sampling process of a GNM by exploiting the DCSD property and iteratively sampling a hierarchy of latent variables that represent cluster activations at different levels..
favorite8For example, we could (1) compactly represent the edge dependencies in the network, and (2) develop more efficient sampling mechanisms based on the conditional independence/dependence relationships encoded in the graphical model structure.
favorite1Some generative network models (models that generate social/information network samples from a network distribution P (G)), with complex interactions among a set of RVs, can be represented with probabilistic graphical models, in particular with BNs. In the present work we show one such a case.
favorite7Context specific independence (CSI) is a property of graphical models where additional independence relations arise in the context of particular values of random variables (RVs).