Nonlinear dynamics in pulse-modulated feedback drug dosing *
Alexander Medvedev, Anton V. Proskurnikov, Zhanybai T. Zhusubaliyev · 2025
Read Nonlinear dynamics in pulse-modulated feedback drug dosing * by Alexander Medvedev, Anton V. Proskurnikov, Zhanybai T. Zhusubaliyev (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-19878.txt.
Introduction
Alexander Medvedev, Anton Proskurnikov and Zhanybai Zhusubaliyev warn that pulse-based drug dosing can behave chaotically, in their 2025 arXiv paper. Many dosing systems deliver discrete pulses rather than a continuous infusion, mimicking manual administration over longer periods. That makes the control loop hybrid: drug dynamics run continuously while the controller acts in steps, and hybrid loops are strongly nonlinear.
Studying anesthesia dosing of a muscle-blocking agent, they find the intended regular rhythm coexists with unwanted periodic patterns, multiple stable regimes and even deterministic chaos. Their conclusion is practical: dosing algorithms built this way need thorough bifurcation analysis before they can be trusted with patient safety.
Catalog record
| CREATOR | Alexander Medvedev, Anton V. Proskurnikov, Zhanybai T. Zhusubaliyev |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 33 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-19878
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Nonlinear dynamics in pulse-modulated feedback drug dosing * Alexander Medvedev, Anton V. Proskurnikov, Zhanybai T. Zhusubaliyev Abstract Pulse-modulated feedback is utilized in drug dosing to mimic sustained over a longer period of time manual discrete dose administration, the latter is in contrast with continuous drug infusion. The intermittent mode of dosing calls for a hybrid (continuous-discrete) modeling of the closed-loop system, where the pharmacokinetics and pharmacodynamics of the drug are captured by differential equations whereas the control law is described by difference equations. Hybrid dynamics are highly nonlinear which complicates formal design of pulse-modulated feedback. This paper demonstrates complex nonlinear dynamical phenomena arising in a simple control system of dosing a neuromuscular blockade agent in anesthesia. Along with the nominal periodic regimen, undesirable nonlinear behaviors, i.e. periodic solutions of high multiplicity, multistability, as well as deterministic chaos, are shown to exist. It is concluded that design of feedback drug dosing algorithms based on a hybrid paradigm has to be informed by a thorough bifurcation analysis in order to secure patient safety. Clinical Relevance —The paper highlights potential patient safety risks posed by complex nonlinear phenomena in closed-loop drug administration. It also presents a systematic approach to mitigate them at the controller design phase. I INTRODUCTION Relieving the burden of manual drug administration over lengthy periods of time requires automation. Surgery with operative time over two hours is common and has to be reliably supported by general anesthesia that allows patients to be unconscious and free of pain throughout the procedure. General anesthesia is nowadays predominantly achieved by intravenous (IV) administration of sedatives (hypnotics), analgesics, and muscle relaxants, i.e. neuromuscular blockade (NMB) agents. Anesthetic drugs are given intravenously either as bolus (one-time) doses or a continuous infusion. A standard procedure in surgery consists of an initial NMB bolus followed by continuous infusion of the drug at a constant flow rate. This mode of administration is supported in target-controlled infusion (TCI), a computerized system that determines the infusion rate required to produce a desired drug concentration at the effect site. In TCI, the required dose is calculated using a pharmacokinetic-pharmacodynamics (PKPD) drug model from patient’s age, sex, weight and other relevant parameters. Programmed intermittent bolus (PIB) is an anesthesia technique in which boluses of an anesthetic drug are automatically injected multiple times, with or without patient-controlled boluses. Being delivered into the epidural space in labor analgesia, PIB has shown to reduce local anesthetic usage and improve maternal satisfaction [ 1 ] . The lack of reliable real-time pain sensing technology [ 2 ] hinders development of a closed-loop version of PIB, where bolus delivery would be controlled by an objective effect measurement. In most cases, NMB agents are administered via continuous infusion. However, controlled boluses are recommended in some cases because intermittent doses allow serial evaluation and reduce the risk of developing myopathies due to prolonged paralysis [ 3 ] . The effect of NMB agents is routinely measured by neuromuscular monitors [ 4 ] , devices that electrically stimulate a peripheral nerve while also quantifying the evoked responses. Compared to the administration of fixed doses (open-loop control), using the monitors for dose titration during the course of treatment significantly reduces the exposure to NMB drugs without affecting the observed clinical outcome [ 5 ] . The present paper investigates the dynamical properties of a drug dosing system that implements PIB administration in a feedback framework. The NMB is selected as application due to the availability of reliable effect quantification that enables feedback control. Using a mixture of analytical and numerical analysis methods, it is demonstrated that complex nonlinear dynamical phenomena can arise in the closed-loop drug administration when continuous effect measurement is utilized to control the regimen of intermittent boluses. The paper is organized as follows. First mathematical models adopted in the study are introduced. Then, the dynamical properties of the closed-loop drug delivery system are summarized. Based on a parsimonious PKPD model of a NMB agent, a pulse-modulated feedback law implementing a clinically established dosing regimen is designed. Further, bifurcation analysis of the closed-loop dynamics is performed to identify complex nonlinear phenomena arising in the drug dosing system operation and illustrate them by simulation. Finally, conclusions are drawn. II Mathematical models Continuous part Consider a time-invariant Wiener system whose measured output is a nonlinear function of the linear block output. The linear block is given by the state-space representation \dot{x}(t)=Ax(t),\quad\bar{y}(t)=Cx(t), (1) where the matrices are A=\begin{bmatrix}-a_{1}&0&0\\ g_{1}&-a_{2}&0\\ 0&g_{2}&-a_{3}\end{bmatrix},C=\begin{bmatrix}0&0&1\end{bmatrix}, (2) a_{1},a_{2},a_{3}>0 are distinct constants, and g_{1},g_{2}>0 are positive gains. The measured output is then y(t)=\varphi(\bar{y}), (3) where \varphi(\cdot) is a smooth function. Discrete part Continuous-time system ( 1 ) is controlled by a pulse-modulated feedback that gives rise to instantaneous jumps in the state vector x(t) \displaystyle x(t_{n}^{+}) \displaystyle=x(t_{n}^{-})+\lambda_{n}B,\quad t_{n+1}=t_{n}+T_{n}, (4) \displaystyle B \displaystyle=\begin{bmatrix}1&0&0\end{bmatrix}^{\top},\quad n=0,1,\ldots, where T_{n}=\bar{\Phi}(y(t_{n})),\;\lambda_{n}=\bar{F}(y(t_{n})). In pulse-modulated control, \bar{\Phi}(\cdot) is referred to as the frequency modulation function and \bar{F}(\cdot) as the amplitude modulation function. The minus and plus in a superscript in ( 4 ) denote the left-sided and right-sided limits, respectively. The described by ( 4 ) control mechanism corresponds to plant ( 1 ) being subject to an impulsive action \lambda_{n}B\delta(t_{n}) applied directly to the state vector, where \delta(\cdot) is Dirac delta function. Despite the jumps in ( 4 ), the output y(t) remains a smooth function since \varphi(\cdot) is continuous and CB=CAB=0,\,CA^{2}B\neq 0. (5) Since both \lambda_{n} and T_{n} are functions of the output value at t_{n} , control law ( 4 ) implements a discrete first-order feedback loop over third-order continuous plant ( 1 ). With \circ denoting composition, introduce the functions \Phi(\cdot)\triangleq(\bar{\Phi}\circ\varphi)(\cdot),\quad F(z)\triangleq(\bar{F}\circ\varphi)(\cdot). Then closed-loop system ( 1 ), ( 4 ) is equivalent to the Impulsive Goodwin’s Oscillator model [ 6 ] , [ 7 ] . It constitutes a hybrid (continuous-discrete) system that is able to exhibit a wide range of nonlinear dynamics phenomena but can also be designed to produce a desired behavior through the choice of the modulation functions \bar{F}(\cdot) , \bar{\Phi}(\cdot) . II-A Pharmacokinetic-pharmacodynamic model A minimally parametrized pharmacokinetic-pharmacodynamic (PKPD) model of the neuromuscular blockade (NMB) agent atracurium is introduced in [ 8 ] . It captures the dose-effect relationship for the drug with a linear dynamical block representing the pharmacokinetics and a nonlinear (static) function describing the pharmacodynamics. Being rewritten in state space, it has the form of ( 1 ), ( 3 ) with a certain choice of the parameters and the output function \varphi(\cdot) . The output y(t) [\%] represents the effect of the NMB agent and is measured by a train-of-four neuromuscular monitor [ 4 ] . The maximal level of y(t)=100\% is achieved when the NMB is initiated and there is no drug in the bloodstream of the patient. The elements of the matrix A , cf. ( 2 ), are parametrized in terms of a common factor \alpha>0 a_{1}\!=\!v_{1}\alpha,a_{2}\!=\!v_{2}\alpha,a_{3}\!=\!v_{3}\alpha,g_{1}\!=\!v_{1}\alpha,g_{2}\!=\!v_{2}v_{3}\alpha^{2}, (6) where v_{1}=1 , v_{2}=4 , v_{3}=10 are fixed coefficients calculated from clinical data. The PD part is modeled by a Hill function of order \gamma y=\varphi(\bar{y})=\frac{100C_{50}^{\gamma}}{C_{50}^{\gamma}+{\bar{y}}^{\gamma}(t)},\quad\gamma>0. (7) Here C_{50}=3.2425 \mu\mathrm{g}\ \mathrm{ml}^{-1} is the drug concentration that produces 50% of the maximum effect. The minimality of the PKPD model consists in its particularization to a patient by means of the pair of constants (\alpha,\gamma)