pubmed.ncbi.nlm.nih.gov

Network Physiology: How Organ Systems Dynamically Interact - PubMed

  • ️Thu Jan 01 2015

Network Physiology: How Organ Systems Dynamically Interact

Ronny P Bartsch et al. PLoS One. 2015.

Abstract

We systematically study how diverse physiologic systems in the human organism dynamically interact and collectively behave to produce distinct physiologic states and functions. This is a fundamental question in the new interdisciplinary field of Network Physiology, and has not been previously explored. Introducing the novel concept of Time Delay Stability (TDS), we develop a computational approach to identify and quantify networks of physiologic interactions from long-term continuous, multi-channel physiological recordings. We also develop a physiologically-motivated visualization framework to map networks of dynamical organ interactions to graphical objects encoded with information about the coupling strength of network links quantified using the TDS measure. Applying a system-wide integrative approach, we identify distinct patterns in the network structure of organ interactions, as well as the frequency bands through which these interactions are mediated. We establish first maps representing physiologic organ network interactions and discover basic rules underlying the complex hierarchical reorganization in physiologic networks with transitions across physiologic states. Our findings demonstrate a direct association between network topology and physiologic function, and provide new insights into understanding how health and distinct physiologic states emerge from networked interactions among nonlinear multi-component complex systems. The presented here investigations are initial steps in building a first atlas of dynamic interactions among organ systems.

PubMed Disclaimer

Conflict of interest statement

Competing Interests: The authors have declared that no competing interests exist.

Figures

Fig 1
Fig 1. Coordinated bursting activities in the output signals from diverse physiological organ systems.

Segments of synchronously recorded physiological signals including (a) brain EEG signal (EEG), (b) heart rate (HR), (c) respiratory rate (Resp) (d) eye movement recording (Eye). Coordinated bursting activities with certain time delay are consistently observed across the output signals of organ systems. Red dashed lines highlight a train of four significant bursts. These bursts transcend all systems and indicate networked communications among the systems.

Fig 2
Fig 2. Schematic presentation of the TDS method.

Segments of brain EEG power spectrum S(f) for the (a) δ band and (b) σ band shown in 60 sec time windows (I), (II), (III) and (IV). Synchronous bursts in S δ and S σ lead to pronounced cross-correlation (shown in (c)) within each time window in (a) and (b), and to a stable time delay characterized by segments of constant τ 0 as shown in (d)—e.g., four red dots highlighted by a grey box in panel (d) represent the time delay in the cross-correlation function peaks for each of the 4 time windows in (c). Note the transition from strongly fluctuating behavior in τ 0 to a stable time delay regime at the transition from deep sleep to light sleep at ∼1200 sec in panel (d). The TDS analysis (See Section Method) is performed on overlapping moving windows with a step of 30 sec. Long periods of constant time delay τ indicate strong TDS coupling represented by strong links in the network of physiologic interactions. The TDS approach is general and can identify and quantify interactions between systems with diverse dynamics and characteristic time scales.

Fig 3
Fig 3. Inter- and intra-channel brain networks during different physiologic states.

Network nodes with different colors represent seven different frequency bands (δ, θ, α, σ, β, γ 1, γ 2) derived from EEG signals (see Section Data), and each set of seven nodes ordered as a heptagon forms a vertex on the hexagon representing six EEG channels from particular brain locations: 2 Frontal areas (Fp1 and Fp2), 2 Central areas (C3 and C4) and 2 Occipital areas (O1 and O2). Coupling strength between frequency bands of signals from different EEG channels (i. e., inter-channel networks) is quantified as the fraction of time (out of the total duration of a given sleep stage throughout the night) when TDS is observed (upper panel). While during quiet W and LS the network of inter-channel brain interactions exhibit high connectivity and strong links between frequency bands of different EEG channels, the networks during REM and DS are more sparse with weaker links. Links between frequency bands of the same EEG channel (i.e., intra-channel networks) are shown in the lower panel. Note that the characteristics of the intra-channel networks for the Frontal areas do not exhibit sleep-stage dependence. In contrast, Central areas exhibit pronounced sleep-stage stratification in network structure, while intra-channel networks in the Occipital areas are more sparse for all sleep stages. Links between two nodes represent the group averaged coupling strength between frequency bands over all subjects. Inter-channel links strength is divided into three categories: very strong links with %TDS≥80% (thick magenta lines); strong links with 65%≤%TDS<80% (thick blue lines); and intermediately strong links with 45%≤%TDS<65% (thin cyan lines). For the intra-channel networks, only links with %TDS≥45% are shown.

Fig 4
Fig 4. Network characteristics of brain-brain interactions of the inter-channel sub-networks.

Pronounced sleep-stage stratification pattern is observed in both (a) number of links and (b) group averaged links strength. Notably, there is a significant difference between LS and DS, as well as between W and REM. This is in contrast to traditional sleep-stage classification of physiologic dynamics of individual systems, where DS and LS are considered similar states grouped as Non-REM state, that is separate from REM and W.

Fig 5
Fig 5. Network structure and links strength of inter-channel brain-brain interactions within each brain hemisphere during different sleep stages.

Inter-channel brain links are separately grouped by physiologic state and specific hemisphere. Each group contains three sets of links: Frontal-Central, Central-Occipital and Frontal-Occipital. Network nodes with different colors represent seven different frequency bands (δ, θ, α, σ, β, γ 1, γ 2), as shown in Fig 3. Inter-channel links strength is divided into three categories: very strong links with %TDS≥80% (thick magenta lines); strong links with 65%≤%TDS<80% (thick blue lines); and intermediately strong links with 45%≤%TDS<65% (thin cyan lines). A very different network structure is associated with each sleep stage, and a hierarchical reorganization involving specific building blocks is observed with transitions across sleep stages (top panels in (a) and (b)). Sleep-stage specific network structure is coupled with pronounced rank order in the links strength across brain areas: strongest links between Frontal and Central areas; intermediate links between Central and Occipital areas and weaker links between Frontal and Occipital areas. Notably, this rank order exhibits pronounced stratification pattern across sleep stages. Both network structure and links strength show a remarkable symmetry between the left and right hemisphere.

Fig 6
Fig 6. Network structure and links strength of inter-channel brain-brain interactions across brain hemispheres during different sleep stages.

Inter-channel brain links are grouped by physiologic state and are separately plotted for: (a) the horizontal links across the left and right brain hemisphere including three subsets: Frontal-Frontal, Central-Central and Occipital-Occipital links; and (b) the diagonal links including six sets of cross-hemisphere links: 2 Frontal-Central, 2 Central-Occipital and 2 Frontal-Occipital links. Network nodes with different colors represent seven different frequency bands (δ, θ, α, σ, β, γ 1, γ 2), as shown in Fig 3. Inter-channel links strength is divided into three categories: very strong links with %TDS≥80% (thick magenta lines); strong links with 65%≤%TDS<80% (thick blue lines); and intermediately strong links with 45%≤%TDS<65% (thin cyan lines). A distinct network structure is associated with each sleep stage ((a) and (b), top panels). Further, a consistent rank order in the group averaged links strength and a pronounced sleep-stage stratification pattern is observed for all cross-hemisphere inter-channel networks (similar to the one observed for the networks within each hemisphere shown in Fig 5), indicating a basic rule of brain-brain network dynamics.

Fig 7
Fig 7. Networks of brain-eye interactions during different physiologic states.

Brain areas are represented by Frontal (Fp1 and Fp2), Central (C3 and C4) and Occipital (O1 and O2) EEG channels. Network nodes with different colors represent seven frequency bands (δ, θ, α, σ, β, γ 1, γ 2) in the spectral power of each EEG channel. Network links between the eye (pink hexagon) and EEG frequency nodes at different locations are determined based on the TDS measure (See Section Methods), and links strength is illustrated by the line thickness. Shown are all links with strength ≥5%TDS. Radar-charts centered in each hexagon represent the relative contribution of brain control from different brain areas to the strength of network links during different sleep stages. The length of each segment along each radius in the radar-charts represents TDS coupling strength between the eye and specific each frequency band at each EEG channel location. These segments are shown in the same color as the corresponding EEG frequency nodes. The brain-eye network interactions are mediated mainly through low-frequency δ bands (thick blue links). Dominant links to the Frontal brain areas are observed for all sleep stages and indicate strong spatial asymmetry in brain-eye network interactions. A network re-organization is observed with transition from one sleep stage to another with a pronounced stratification pattern in the overall strength of network links—higher strength in W and LS (larger hexagons) and weaker links during REM and DS (smaller hexagons).

Fig 8
Fig 8. Histograms of links strength in the brain-eye network during different sleep stages.

Group averaged links strength are obtained using the TDS measure, where each link represents the interaction of the eye with a given brain area through a specific frequency band. Links are separately grouped by brain areas (Frontal, Central or Occipital) and are arranged in order from low-frequency (δ and θ) to high-frequency (γ 1 and γ 2) bands. A consistent frequency profile of links strength is observed for brain areas—strongest links mediated through the low-frequency δ band, and weak links mediated through the mid-range frequency bands. Links strength between the eye and different brain areas decreases from the Frontal to Central and Occipital areas. The sleep-stage stratification observed for the links strength in the brain-eye radar-charts (Fig 7) is consistent for all frequency bands and brain areas as shown in the histograms. Note the strong symmetry in the links strength distribution between the left and right hemisphere that is present for all sleep stages.

Fig 9
Fig 9. Networks of brain-chin interactions during different physiologic states.

Brain areas are represented by Frontal (Fp1 and Fp2), Central (C3 and C4) and Occipital (O1 and O2) EEG channels. Network nodes with different colors represent seven frequency bands (δ, θ, α, σ, β, γ 1, γ 2) in the spectral power of each EEG channel. Network links between the chin (orange hexagon) and EEG frequency nodes at different locations are determined based on the TDS measure (See Section Methods), and links strength is illustrated by the line thickness. Shown are all links with strength ≥5%TDS. Radar-charts centered in each hexagon represent the relative contribution of brain control from different brain areas to the strength of network links during different sleep stages. The length of each segment along each radius in the radar-charts represents TDS coupling strength between the chin and each frequency band at each EEG channel location. These segments are shown in the same color as the corresponding EEG frequency nodes. The brain-chin network interactions are mediated mainly through high-frequency γ 1 and γ 2 bands (orange and red links). In contrast to the brain-eye network (Fig 7), the brain-chin network is characterized by relatively symmetric links strength to all six brain areas, as shown by the symmetric radar-chart in each hexagon. A pronounced stratification pattern is observed for the overall strength of network links—stronger links during W and LS (larger hexagons) and weaker links during REM and DS (smaller hexagons).

Fig 10
Fig 10. Histograms of links strength in the brain-chin network during different sleep stages.

Group averaged links strength are obtained using the TDS measure, where each link represents the interaction of the chin with a given brain area through a specific frequency band. Links are separately grouped by brain areas (Frontal, Central or Occipital), and are arranged in order from low-frequency (δ and θ) to high-frequency (γ 1 and γ 2) bands. A consistent frequency profile of links strength is observed for all brain areas—strongest links mediated through the highest-frequency γ 2 band, a gradual decrease in link strength for the lower-frequency bands (γ 1, β, σ, α, θ), followed by a slight increase in link strength for the lowest-frequency δ band. The sleep-stage stratification observed for the links strength in the brain-chin radar-charts (Fig 9) is consistently observed for all frequency bands and brain areas as shown in the histograms—stronger links during W and LS, and weaker links during REM and DS. Note the strong symmetry in the links strength distribution between the left and right hemisphere that is present for all sleep stages.

Fig 11
Fig 11. Networks of brain-leg interactions during different physiologic states.

Brain areas are represented by Frontal (Fp1 and Fp2), Central (C3 and C4) and Occipital (O1 and O2) EEG channels. Network nodes with different colors represent seven frequency bands (δ, θ, α, σ, β, γ 1, γ 2) in the spectral power of each EEG channel. Network links between the leg (dark green hexagon) and EEG frequency nodes at different locations are determined based on the TDS measure (See Section Methods), and links strength is illustrated by the line thickness. Shown are links with strength ≥5%TDS. Radar-charts centered in each hexagon represent the relative contribution of brain control from different brain areas to the strength of network links during different sleep stages. The length of each segment along each radius in the radar-charts represents TDS coupling strength between the leg and each frequency band at each EEG channel location. These segments are shown in the same color as the corresponding EEG frequency nodes. The brain-leg network interactions are mediated mainly through high-frequency γ 2 bands (red links). A pronounced stratification pattern is observed for the overall strength of network links—stronger links during W (larger hexagon), intermediate during REM and LS (smaller hexagon) and weaker links during DS. Since the same scaling factor is used to obtain the radar-charts for brain-leg network as for the brain-chin network (Fig 9), smaller radar-charts (smaller hexagons) indicate weaker interactions in the brain-leg network compared to the brain-chin network.

Fig 12
Fig 12. Histograms of links strength in the brain-leg network during different sleep stages.

Group averaged links strength are obtained using the TDS measure, where each link represents the interaction of the leg with a given brain area through a specific frequency band. Links are separately grouped by brain areas (Frontal, Central or Occipital), and are arranged in order from low- (δ and θ) to high-frequency (γ 1 and γ 2) bands. A consistent frequency profile of links strength is observed for all brain areas—strongest links mediated through the highest frequency γ 2 band, followed by a gradual decrease in link strength for the lower-frequency bands (γ 1, β, σ, α, θ, δ). The sleep-stage stratification observed for the links strength in the brain-leg radar-charts (Fig 11) is consistent for all frequency bands and brain areas as shown in the histograms. Note the strong symmetry in the links strength distribution between the left and right hemisphere that is present for all sleep stages.

Fig 13
Fig 13. Networks of brain-heart interactions during different physiologic states.

Brain areas are represented by Frontal (Fp1 and Fp2), Central (C3 and C4) and Occipital (O1 and O2) EEG channels. Network nodes with different colors represent seven frequency bands (δ, θ, α, σ, β, γ 1, γ 2) in the spectral power of each EEG channel. Network links between the heart (red hexagon) and EEG frequency nodes at different locations are determined based on the TDS measure (See Section Methods), and links strength is illustrated by the line thickness. Shown are links with strength ≥5%TDS. Radar-charts centered in each hexagon represent the relative contribution of brain control from different brain areas to the strength of network links during different sleep stages. The length of each segment along each radius in the radar-charts represents TDS coupling strength between the heart and each frequency band at each EEG channel location. These segments are shown in the same color as the corresponding EEG frequency nodes. During W and REM, the brain-heart network interactions are mediated mainly through high-frequency γ 1 and γ 2 bands (orange and red links), while during LS and DS, the interactions are mediated uniformly through all frequency bands. In contrast to the brain-eye network (Fig 7), the brain-heart network is characterized by relatively symmetric links strength to all six brain areas, as shown by the symmetric radar-charts in each hexagon. A pronounced stratification pattern is observed for the overall strength of network links—stronger links during W and LS (larger hexagons) and weaker links during REM and DS (smaller hexagons). Notably, compared to the brain-eye (Fig 7), brain-chin (Fig 9) and brain-leg networks (Fig 11), there are no links in the brain-heart network during DS (all links <5%TDS).

Fig 14
Fig 14. Histograms of links strength in the brain-heart network during different sleep stages.

Group averaged links strength are obtained using the TDS measure, where each link represents the interaction of the heart with a given brain area through a specific frequency band. Links are separately grouped by brain areas (Frontal, Central or Occipital), and are arranged in order from low-frequency (δ and θ) to high-frequency (γ 1 and γ 2) bands. For a given physiologic state, the frequency profile of the TDS links strength of brain-heart interactions remains stable for all brain areas. In contrast to other brain-organ networks where the frequency profile is stable across sleep stages, the frequency profiles of links strength in the brain-heart network change with transitions across different physiologic states—for all brain areas the frequency profiles become flatter as sleep depth increases—indicating that the mechanisms driving brain-heart interactions are markedly different compared to other organs. The sleep-stage stratification observed for the links strength in the brain-heart radar-charts is consistently observed for all frequency bands and brain areas as shown in the histograms—stronger links during W and LS, and weaker links during REM and DS. There is a strong symmetry in the links strength distribution between the left and right hemispheres during all sleep stages.

Fig 15
Fig 15. Networks of brain-respiration interactions during different physiologic states.

Brain areas are represented by Frontal (Fp1 and Fp2), Central (C3 and C4) and Occipital (O1 and O2) EEG channels. Network nodes with different colors represent seven frequency bands (δ, θ, α, σ, β, γ 1, γ 2) in the spectral power of each EEG channel. Network links between the respiration (purple hexagon) and EEG frequency nodes at different locations are determined based on the TDS measure (See Section Methods), and links strength is illustrated by the line thickness. Shown are links with strength ≥3%TDS. Radar-charts centered in each hexagon represent the relative contribution of brain control from different brain areas to the strength of network links during different sleep stages. The length of each segment along each radius in the radar-charts represents TDS coupling strength between respiration and each frequency band at each EEG channel location. These segments are shown in the same color as the corresponding EEG frequency nodes. A pronounced stratification pattern is observed for the overall strength of network links—stronger links during W and LS (larger hexagons) and weaker links during REM and DS (smaller hexagons). Note the much weaker links (overall smaller hexagons) for all sleep stages in the brain-respiration network compared to other brain-organ networks, indicating much weaker coupling between the brain and respiration system as measured by %TDS. Similar to the brain-heart network (Fig 13), there are no links in the brain-respiration network during DS (all links <3%TDS).

Fig 16
Fig 16. Histograms of links strength in the brain-respiration network during different sleep stages.

Group averaged links strength are obtained using the TDS measure, where each link represents the interaction of the respiratory system with a given brain area through a specific frequency band. Links are separately grouped by brain areas (Frontal, Central or Occipital), and are arranged in order from low-frequency (δ and θ) to high-frequency (γ 1 and γ 2) bands. Brain-respiration networks are characterized by a very homogeneous frequency profile of links strength which is consistently observed across all brain areas—an almost flat distribution across all 7 physiologically relevant frequency bands (δ, θ, α, σ, β, γ 1, γ 2). The sleep-stage stratification observed for the links strength in the brain-respiration radar-charts (Fig 15) is consistently observed for all frequency bands and brain areas although less pronounce compared to other brain-organ networks. A strong symmetry in the links strength distribution between the left and right hemisphere is present for all sleep stages. Overall brain-respiration links are much weaker than in other brain-organ networks.

Fig 17
Fig 17. Networks of physiologic interactions between brain areas and key organ systems during different physiologic states.

Brain areas are represented by Frontal (Fp1 and Fp2), Central (C3 and C4) and Occipital (O1 and O2) EEG channels. Interactions between brain channels and organ systems are represented by weighted undirected graphs. The size of each organ node in the network is proportional to the strength of the overall brain-organ interaction as measured by the summation of the TDS links strength for all frequency bands and EEG channel locations. The color of each organ node corresponds to the dominant frequency band in the coupling of the organ system with the brain. The width of each link reflects the strength of dynamic coupling as measured by %TDS, and colors of the links correspond to the colors of the nodes representing the different frequency bands (color bars). Plotted are only links with strength ≥3%TDS. Thicker links correspond to stronger coupling and higher time delay stability. The physiological network exhibits transitions across sleep stages—lowest number of links during DS, higher during REM, and highest during LS and W. For different organs, brain-organ interactions are mediated through different dominant frequency bands, e.g., the chin and the leg are predominantly coupled to the brain through the high-frequency γ 2 band during all sleep stages whereas brain-eye network interactions are mediated mainly through low-frequency δ band. The complex networks of dynamic interactions between key organ systems and the brain undergoes a hierarchical reorganization across different sleep stages, indicating a previously unknown mechanism of regulation.

Fig 18
Fig 18. Networks of physiologic interactions among key organ systems during different physiologic states.

Interactions among organ systems are represented by weighted undirected graphs, where links reflect the strength of dynamic coupling as measured by % TDS (Section Methods). Darker and thicker links between organ systems correspond to stronger interaction with higher %TDS. The size of each organ node in the network is proportional to the strength of the overall brain-organ interaction as measured by the summation of the TDS links strength for all frequency bands and EEG channel locations. Hexagons representing individual organs in the networks are obtained in the same way as in Figs 7, 9, 11, 13, and 15; and are normalized to the same size. Color bar represents different physiologically relevant frequency bands in the EEG spectral power and is used in the radar-charts for the brain-organ interactions shown in each hexagon. The color of each organ node as well as the edge color of the organ hexagon corresponds to the dominant frequency band in the coupling of the organ system with the brain. Notably, larger organ nodes representing stronger brain-organ interactions are consistently connected by stronger organ-organ links (thicker and darker lines). A pronounced re-organization in the configuration of network links strength is observed with transitions from one sleep stage to another, demonstrating a clear association between network structure and physiologic function.

Similar articles

Cited by

References

    1. Bashan A, Bartsch RP, Kantelhardt JW, Havlin S, Ivanov PCh. Network physiology reveals relations between network topology and physiological function. Nat Commun. 2012;3:702 10.1038/ncomms1705 - DOI - PMC - PubMed
    1. Ivanov PCh, Bartsch RP. Network Physiology: Mapping Interactions Between Networks of Physiologic Networks In: D’Agostino G, Scala A, editors. Networks of Networks: The Last Frontier of Complexity. Springer International Publishing; 2014. p. 203–222.
    1. Quill E. When Networks Network. Science News. 2012;182(6):18–25. 10.1002/scin.5591820619 - DOI
    1. Arodz T, Bonchev D. Identifying influential nodes in a wound healing-related network of biological processes using mean first-passage time. New Journal of Physics. 2015;17(2):025002 10.1088/1367-2630/17/2/025002 - DOI
    1. Rothkegel A, Lehnertz K. Irregular macroscopic dynamics due to chimera states in small-world networks of pulse-coupled oscillators. New Journal of Physics. 2014;16(5):055006 10.1088/1367-2630/16/5/055006 - DOI

Publication types

MeSH terms

LinkOut - more resources