A semi‐adaptive control approach to closed‐loop medication infusion

Abstract

This paper presents a semi-adaptive control approach to closed-loop medication infusion problems. The rationale underlying this approach is to design a controller that can adapt model parameters with a large impact on the model's fidelity while fixing the remaining parameters at nominal values. In this paper, a control-oriented model for this purpose is derived via system identification and sensitivity analysis of a loworder model capturing the direct dose-response relationship Using the model thus derived, a model-reference adaptive controller and a composite adaptive controller are designed and compared with each other. In-silico simulation results using remifentanil's effect on respiratory rate as an example indicate that both controllers can regulate the output at commanded set points.

Coordinated semi‐adaptive closed‐loop control for infusion of two interacting medications

“…The steady-state error associated with cardiac output was relatively large with the coordinated controller. This was due to the dead zone implemented for the parameter adaptation law in (16). In the absence of coordinated controller, the adaptation law was persistently enabled to drive cardiac output error to zero because the tracking error associated with respiratory rate was large (since its reference target could not be reached) and the dead zone was never activated.…”

“…For the parameters associated with the dose-response relationships for remifentanil, we derived the ranges of k e2 , I 50, 22 , and γ 2 from our previous work, 16 while we assumed that the influence of remifentanil on cardiac output is relatively small (translating to a large I 50, 12 ). The selected ranges for the parameters were 0.10 ≤ k e2 ≤ 0.50 minutes −1 , 1 ≤ γ 2 ≤ 5, 0.12 ≤ I 50, 12 ≤ 0.36 mcg/kg/min, and 0.04 ≤ I 50, 22 ≤ 0.12 mcg/kg/min.…”

“…We developed a dose-response model for 2 interacting medications by combining a low-order mixing model reported in our previous works 16,26 and a response surface model reported in Minto et al 27 (see Figure 1). First, the low-order mixing model represents the relationship between the intravenous infusion rate and the (hypothetical) infusion rate at the site of action of a medication and is described as follows: where u 1 and u 2 are the intravenous infusion rates associated with the 2 medications M 1 and M 2 , x 1 and x 2 are the corresponding infusion rates at the sites of action, and k e1 and k e2 are the equilibration constants.…”

“…In a previous work, we have shown that γ i makes relatively small influence on the system output compared with k ei and I 50, ij (i, j = 1, 2) and may thus be fixed at a nominal value. 16 With γ 1 and γ 2 specified a priori, the model (9) is a linearly parameterized control design model with the elements of matrices A and B as unknowns to be adapted online. As the proposed control approach adapts only a subset of the plant model parameters (ie, parameters other than γ 1 and γ 2 ), it is semi-adaptive rather than fully adaptive.…”

“…However, the majority of prior art has focused on the infusion of a single medication, eg, propofol, [13][14][15] remifentanil, 16,17 and fluid, 18,19 whereas there is only limited volume of work reported on the closed-loop control for infusion of multiple medications to track multiple reference targets (see, for example, previous works [20][21][22] ). In addition to closed-loop control design, the coordination of reference targets also presents a challenge.…”