Delay-Induced Uncertainty in Physiological Systems

Bhargav R. Karamched, George Hripcsak, David J. Albers, William OttarXiv 2020 · arXiv

Abstract

Medical practice in the intensive care unit is based on the supposition that physiological systems such as the human glucose-insulin system are reliable. Reliability of dynamical systems refers to response to perturbation: A dynamical system is reliable if it behaves predictably following a perturbation. Here, we demonstrate that reliability fails for an archetypal physiological model, the Ultradian glucose-insulin model. Reliability failure arises because of the presence of delay. Using the theory of rank one maps from smooth dynamical systems, we precisely explain the nature of the resulting delay-induced uncertainty (DIU). We develop a recipe one may use to diagnose DIU in a general dynamical system. Guided by this recipe, we analyze DIU emergence first in a classical linear shear flow model and then in the Ultradian model. Our results potentially apply to a broad class of physiological systems that involve delay.

Introduction

In clinical settings, particularly the Intensive Care Unit (ICU), medical practitioners operate under the assumption of reliability. This concept implies that if a physiological system is perturbed—for instance, by administering an insulin shot—the resulting response should depend predictably on the intervention itself, rather than the microscopic state of the patient at that specific millisecond.

However, many physiological processes involve significant time delays, such as the time required for hormone secretion, circulation, and receptor binding. This research explores whether these delays can cause a "reliability failure," termed Delay-Induced Uncertainty (DIU). In such states, the timing of a medical intervention becomes so sensitive that identical doses could lead to wildly different, erratic outcomes.

Methodology and Models

The researchers investigated DIU using a combination of mathematical theory and numerical simulation. They proposed a three-part "recipe" for identifying reliability loss:

  1. Excitability: Delay makes the system excitable, often creating a weakly stable limit cycle (oscillation).
  2. Intrinsic Shear: The system exhibits velocity gradients near this cycle.
  3. External Forcing: Periodic or random interventions (e.g., meals) interact with this shear to stretch and fold the phase space.

The Ultradian Glucose-Insulin Model

To test this theory, the authors applied it to the Ultradian glucose-insulin model, a six-dimensional compartment model describing the interactions between glucose and insulin.

The Ultradian glucose-insulin model structure, showing interactions between plasma insulin, interstitial insulin, and plasma glucose.

The model utilizes established parameters derived from clinical studies, representing nominal human physiological values for degradation rates and insulin secretion constants. The primary diagnostic tool used was the maximal Lyapunov exponent ($\Lambda_{max}$); a positive value indicates chaos (DIU), while a negative value indicates reliability.

Key Insights: Delay as a Catalyst for Chaos

The study reveals that delay does more than just slow a system down; it creates structural vulnerabilities. Through the lens of "rank one maps," the paper explains how internal shear causes the state space to wrap around itself when "kicked" by an intervention. If the relaxation time between kicks is sufficient, this process creates a strange attractor—the hallmark of mathematical chaos.

In a DIU state, the timing of a meal or insulin dose can be more critical than the amount. Two patients with identical physiological parameters could exhibit divergent responses to the same treatment based solely on their position in their internal cycle at the time of administration.

Key Results

Numerical experiments confirmed that the Ultradian model enters a chaotic state when subjected to periodic nutritional "kicks," provided the delay ($\tau$) and inter-kick time ($T$) are sufficiently large.

Transition to chaos (DIU) in the Ultradian model. Panels (a-d) show glucose time series for increasing delays, while (e-h) show the corresponding phase-space trajectories transitioning to a strange attractor.

As shown in the figure above, $\Lambda_{max}$ transitions from negative to positive as the inter-kick interval increased, signaling the onset of DIU. Furthermore, DIU was found to be even more robust when "kicks" occurred at random (Poissonian) intervals, mimicking real-world behavior more closely than strictly periodic forcing.

Limitations and Future Directions

While this research provides a rigorous framework for DIU, several challenges remain:

  • Nonstationarity: Real-world patients are "nonstationary," with underlying physiology that shifts as they recover or decline.
  • Clinical Detection: We currently lack data-driven tools to detect if a specific patient is in a DIU state in real-time.
  • Potential Benefits: Interestingly, inducing a chaotic state might occasionally stabilize a system's variance, a counter-intuitive possibility for future study.

Ultimately, these results suggest that DIU may be a widespread phenomenon in physiological systems involving feedback delays, necessitating a shift in how we approach predictable treatment in clinical environments.

Summary updated

Abstract

Medical practice in the intensive care unit is based on the supposition that physiological systems such as the human glucose-insulin system are reliable. Reliability of dynamical systems refers to response to perturbation: A dynamical system is reliable if it behaves predictably following a perturbation. Here, we demonstrate that reliability fails for an archetypal physiological model, the Ultradian glucose-insulin model. Reliability failure arises because of the presence of delay. Using the theory of rank one maps from smooth dynamical systems, we precisely explain the nature of the resulting delay-induced uncertainty (DIU). We develop a recipe one may use to diagnose DIU in a general dynamical system. Guided by this recipe, we analyze DIU emergence first in a classical linear shear flow model and then in the Ultradian model. Our results potentially apply to a broad class of physiological systems that involve delay.
BibTeX
@misc{karamched2020delay,
  author = {Bhargav R. Karamched and George Hripcsak and David J. Albers and William Ott},
  title = {{Delay-Induced Uncertainty in Physiological Systems}},
  year = {2020},
  eprint = {2007.09309},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2007.09309},
}
Figure