Frontiers in network physiology最新文献

Introduction: In this work, we introduce a novel approach to one of the historically fundamental questions in neural networks: how to encode information? More particularly, we look at temporal coding in spiking networks, where the timing of a spike as opposed to the frequency, determines the information content. In contrast to previous temporal-coding schemes, which rely on the statistical properties of populations of neurons and connections, we employ a novel synaptic plasticity mechanism that allows the timing to be learnt at the single-synapse level.

Methods: Using a formal basis from information theory, we show how a phase-coded spike train (relative to some 'reference' phase) can, in fact, multiplex multiple different information signals onto the same spike train, significantly improving overall information capacity. We furthermore derive limits on the channel capacity in the phase-coded spiking case, and show that the learning rule also has a continuous derivative in the input-output relation, making it potentially amenable to classical learning rules from artificial neural networks such as backpropagation.

Results: Using a simple demonstration network, we show the multiplexing of different signals onto the same connection, and demonstrate that different synapses indeed can adapt using this learning rule, to specialise to different interspike intervals (i.e., phase relationships). The overall approach allows for denser encoding, and thus energy efficiency, in neural networks for complex tasks, allowing smaller and more compact networks to achieve combinations of tasks which traditionally would have required high-dimensional embeddings.

Discussion: Although carried out as a study in computational spiking neural networks, the results may have insights for functional neuroscience, and suggest links to mechanisms that have been shown from neuroscientific studies to support temporal coding. To the best of our knowledge, this is the first study to solve one of the outstanding problems in spiking neural networks: to demonstrate that distinct temporal codings can be distinguished through synaptic learning.

In this paper, we calculate magnitude-constrained optimal stimuli for desynchronizing a population of neurons by maximizing the Lyapunov exponent for the phase difference between pairs of neurons while simultaneously minimizing the energy which is used. This theoretical result informs the way optimal inputs can be designed for deep brain stimulation in cases where there is a biological or electronic constraint on the amount of current that can be applied. By exploring a range of parameter values, we characterize how the constraint magnitude affects the Lyapunov exponent and energy usage. Finally, we demonstrate the efficacy of this approach by considering a computational model for a population of neurons with repeated event-triggered optimal inputs.

This paper refers to an issue Haken and myself were discussing, started to work on, prepared a preliminary draft, but never managed to complete and transform it into a full-scale study and publication. Here, in memoriam of Hermann Haken, my dear friend and colleague for many years, I present it as it is - an unfinished study with some innovative ideas that will have to be further elaborated in the future.

Introduction: In previous research on reentrant atrial tachycardia (AT), the index theorem has proven instrumental in uncovering consistent paired counter-rotating anatomical reentry (either complete or near-complete), driving the arrhythmia rotating around critical boundaries (CB). Furthermore, interconnecting each CB-pair with an ablation line has been shown to terminate the AT. In this study, we extend this approach to scar-related ventricular tachycardia (VT), complicating the calculations as VT is inherently a 3-dimensional problem. We propose that scar-related VT can be topologically simplified to one or more of four basic physiologically distinct scar-types: transmural (I-shaped), epicardially connected or endocardially connected (U-shaped) or intramural (O-shaped).

Methods: Six simulations of scar-related VT were created, each featuring a distinct critical scar configuration. From each simulation, three transmural layers (endocardium, mid-myocardium and epicardium) were extracted to create 2-dimensional surfaces, which were analyzed with the index theorem, using the software package Directed Graph Mapping (DGM) extended with novel algorithms to detect the CBs.

Results: On each layer, either no CBs were found or pairs of counter-rotating CBs were found, each CB had an opposite sign, adhering to the index theorem. Ablation was performed by connecting each pair of counter-rotating CBs on each layer to form a continuous ablation surface, bounded by scar tissue, the endocardial surface, or the epicardial surface. This ablation strategy consistently terminated all simulations, supporting the applicability of our topology-based approach to VT.

Conclusion: The index theorem remains valid for scar-related VT. Successful ablation on VT should include, connecting the CB-pairs in each 2 dimensional surface. Any other type of ablation does not terminate the VT.

Deep brain stimulation (DBS) at high frequencies has revolutionized efforts to alleviate Parkinson's disease symptoms for approximately 30 years. Since then, there has been vast investigation into the mechanisms of action of DBS. Recently, synaptic suppression was found to play a pivotal role in the fundamental mechanisms underlying DBS. Based on this understanding, researchers introduced two novel DBS pulsing strategies that use a minimal number of stimuli. In contrast to conventional DBS (cDBS) pulsing, which employs continuous high-frequency pulses (>100 Hz), the two novel methods incorporate changes in pulsing frequency and on/off pulsing periods. In this computational study, we investigated the network effects of these two suggested patterns using an updated version of a biophysically realistic thalamocortical network model of DBS. Both suggested pulsing patterns significantly reduced the exaggerated beta power (∼13 Hz-30 Hz oscillations) in the motor cortex, with careful consideration of the intensity of the stimulating pulses. In addition, they significantly reduced the level of network synchronization. We compared these findings with the effects of 20 and 130 Hz cDBS on our network model and did not observe effects contrary to those of 130 Hz cDBS. The two suggested patterns, which were computationally successful in reproducing known DBS network effects, could potentially increase the battery life of DBS device and reduce the microlesion effect associated with long-term cDBS pulsing. These outcomes, however, require confirmation in further studies.

Conventionally it is assumed that the nerve impulse is an electrical process based upon the observation that electrical stimuli produce an action potential as defined by Hodgkin Huxley (1952) (HH). Consequently, investigations into the computation of nerve impulses have almost universally been directed to electrically observed phenomenon. However, models of computation are fundamentally flawed and assume that an undiscovered timing system exists within the nervous system. In our view it is synchronisation of the action potential pulse (APPulse) that effects computation. The APPulse, a soliton pulse, is a novel purveyor of computation and is a quantum mechanical pulse: i.e., It is a non-Turing synchronised computational event. Furthermore, the APPulse computational interactions change frequencies measured in microseconds, rather than milliseconds, producing effective efficient computation. However, the HH action potential is a necessary component for entropy equilibrium, providing energy to open ion channels, but it is too slow to be functionally computational in a neural network. Here, we demonstrate that only quantum non-electrical soliton pulses converging to points of computation are the main computational structure with synaptic transmission occurring at slower millisecond speeds. Thus, the APPulse accompanying the action potential is the purveyor of computation; a novel computational mechanism, that is incompatible with Turing timed computation and artificial intelligence (AI).

Representing the brain as a complex network typically involves approximations of both biological detail and network structure. Here, we discuss the sort of biological detail that may improve network models of brain activity and, conversely, how standard network structure may be refined to more directly address additional neural properties. It is argued that generalised structures face the same fundamental issues related to intrinsicality, universality and functional meaningfulness of standard network models. Ultimately finding the appropriate level of biological and network detail will require understanding how given network structure can perform specific functions, but also a better characterisation of neurophysiological stylised facts and of the structure-dynamics-function relationship.

Introduction: Parasympathetic nervous system (PNS) dysfunction in Parkinson's disease (PD) has been frequently evaluated using heart rate variability (HRV) analysis in the time and frequency domains. Findings across studies have been inconsistent, limiting a unified understanding of early autonomic impairment.

Methods: In this study, we applied both conventional and advanced analytical methods to evaluate cardiovascular PNS function in the early-stage PD patients. Sixteen individuals with PD (<6 months after motor signs occurrence) and sixteen age- and sex-matched healthy controls were assessed across three protocol phases (supine rest, head-up tilt, and supine recovery). Traditional HRV analysis in the high-frequency band was used to estimate the overall respiratory heart rate variability (RespHRV; updated and more appropriate term for the respiration-related heart rate oscillations formerly called respiratory sinus arrhythmia, RSA) magnitude. To distinguish between baroreflex-mediated and non-baroreflex RespHRV mechanisms, we employed multiscale Partial Information Decomposition (PID), an information-theoretic method. Cardiac baroreflex sensitivity (BRS), reflecting reflex parasympathetic control, was assessed using a causal estimation approach, further supported by a PID-derived parameter quantifying coupling between systolic arterial pressure and R-R intervals. Additionally, the presence of constipation - a clinically relevant non-motor symptom indicative of parasympathetic dysfunction was used to stratify the PD cohort.

Results: Early-stage PD patients exhibited signs of parasympathetic impairment, particularly during orthostatic stress. HRV analysis showed reduced HF power during head-up tilt, while causal BRS was significantly lower across all protocol phases in the PD group. PID analysis further demonstrated a significant reduction in baroreflex-mediated mechanism of RespHRV during head-up tilt in PD patients compared with healthy controls, indicating early dysfunction of the cardiac chronotropic baroreflex. This impairment was more pronounced in the group with gastrointestinal issues (with the presence of constipation).

Discussion: These findings support the α-Synuclein Origin site and Connectome model, which proposes that PD patients whose neurodegeneration originates in the peripheral autonomic nervous system are characterized by early and more severe autonomic dysfunction.

Introduction: During exercise, the cardiovascular, respiratory, and locomotor systems interplay dynamically, yet the specific mechanisms of cardiovascular and locomotor interaction during simple rhythmic exercise like walking remain unclear. Computational models constitute a powerful tool to investigate the interplay of networked physiological systems, but while gravitational and postural effects on circulation have been explored, the influence of inertial forces from body motion on hemodynamics has not been addressed.

Methods: Here, we present a closed-loop cardiovascular model that incorporates inertial effects during walking. The lumped parameter model includes 25 vascular compartments, a four-chamber heart with valves, pericardial and intrathoracic pressures, interventricular septal dynamics, and a baroreflex mechanism. Inertial effects are modeled as additional hydrodynamic pressure sources in each vascular segment, equivalent to the acceleration of blood mass, caused by gravity and motion. Three protocols are used: a head-up tilt test to validate baroreflex and gravity effects; a synthetic walking simulation with controlled heart rate (HR) and step rate (SR); and a human walking experiment (n=2) linking beat-wise simulated aortic pressure to measured brachial pressure using recorded HR and body acceleration. Beat-wise morphology similarity (K-stat) between experimental and simulated hemodynamic waveforms is quantified with a two-sample Kolmogorov-Smirnov test.

Results: The model reproduces expected physiological responses to head-up tilt. During synthetic walking, inertial effects result in pressure augmentation, increasing systolic or diastolic pressure depending on the phase between HR and SR. With SR > HR, phase variability produces a low-frequency "beating" in the pressure waveforms and mean arterial pressure, corresponding to the difference between SR and HR. In the human subject experiment, the model accurately replicates beat-wise pressure changes at varying phase shifts between HR and SR. Quantitative comparison shows a substantial increase in similarity of waveform when hydrodynamic pressure is included (K-stat: 0.123 vs. 0.029 for P1; 0.164 vs. 0.059 for P2).

Conclusion: Introducing contributions of body acceleration as an additional hydrodynamic pressure source in the vascular compartments seems a valid way to capture walking-induced inertial effects. This work contributes to the broader effort to characterize physiological network adaptations to exercise and offers a foundation for future research studying and optimizing cardiac-locomotor interaction.

Introduction: Major depressive disorder (MDD) and MDD with suicidal ideation (MDDSI) present with heterogeneous symptoms, complicating diagnosis and treatment. Precision psychiatry addresses this challenge by applying computational methods and digital biomarkers to objectively distinguish psychiatric states. While psychiatric research has traditionally focused on neural activity, increasing evidence highlights the value of autonomic indices, particularly heart rate variability (HRV), in capturing clinically relevant dysregulation. Cardio-respiratory coupling (CRC), which reflects bidirectional interactions between cardiovascular and respiratory systems, represents a physiologically grounded extension of this approach. Although less frequently applied in psychiatry compared to HRV, CRC offers a sensitive window into autonomic network dynamics and holds promise for differentiating between MDD and MDDSI.

Methods: A total of 74 participants were assigned to Control (n = 35), MDD (n = 21), or MDDSI (n = 18) groups. ECG, PPG, and respiratory signals were recorded at rest and segmented into 2-min intervals. Swarm Decomposition (SwD) was applied to extract four oscillatory components (OC1-OC4) from each signal that go from low to high frequency, respectively. Fractal dimension (Higuchi, Katz) and Shannon entropy quantified coupling complexity. Bidirectional (λbi) and unidirectional (λ) coupling measures and phase angles were computed between respiratory signals and cardiovascular markers: pulse wave amplitude (PWA), pulse transit time (PTT), and pulse rate (PR). Group differences were evaluated using Kruskal-Wallis and post hoc tests (p < 0.05).

Results: Bidirectional PR coupling in OC3 showed significant group differences (p < 0.01). Higuchi fractal dimension of PTT in OC3 was reduced in MDDSI compared to MDD and controls (p = 0.018), suggesting diminished complexity. For PWA in OC4, high-frequency power significantly differed between controls and MDDSI (p = 0.004). Directional coupling entropy also distinguished MDD from MDDSI (p = 0.039).

Conclusion: This study reveals that frequency-specific disruptions in bidirectional cardiorespiratory coupling, along with reduced signal complexity and entropy, are characteristic of MDDSI. These features may reflect impaired autonomic adaptability and emotional regulation. Phase-based coupling metrics and SwD show promise as physiological biomarkers for early identification of high-risk depressive states in digital psychiatry.