Dynamical networks of integrated physiologic systems: network transitions across physiologic states

Tell a Friend
Ronny Bartsch, Harvard Medical School
03/11/2013 - 14:00
Resnick Building (209), seminar room 210

The human organism is an integrated complex network of interconnected and interacting organ  systems, where the behavior of one system may affect the dynamics of all other systems.  Due to these interactions, failure of one system can trigger a breakdown of the entire network.  We introduce a systematic method to identify a network of interactions between diverse physiologic  systems, to quantify the hierarchical structure and dynamics of this network, and to track its  evolution under different physiologic states. We find a robust relation between network structure  and physiologic states: every state is characterized by specific network topology, node connectivity  and links strength -- a behavior we consistently observe across individual subjects. Further, we find  that transitions from one physiologic state to another trigger a markedly fast reorganization of physiologic
interactions on time scales of just a few minutes, indicating high network flexibility in response to  perturbations. Surprisingly, this reorganization occurs simultaneously and globally in the entire  network as well as at the level of individual network nodes, while preserving a hierarchical order in  the strength of network links. In the context of sleep-stage transitions, we demonstrate that  network connectivity and overall strength of physiologic interactions are significantly higher during  wake and light sleep, intermediate during rapid eye movement (REM) sleep and much lower during  deep sleep -- a stable stratification pattern which indicates that physiologic systems are highly and
strongly connected during light sleep, and practically disconnected during deep sleep. Such
pronounced difference in network organization during light and deep sleep is in contrast to the similarity  in the output dynamics of individual physiologic systems during these sleep stages. Our findings  highlight the need of an integrated network approach to understand physiologic function, since the framework we develop provides new information which can not be obtained by studying individual systems.