(448d) Design of a Drug Infusion Controller for Reducing Post-Operative High Blood Pressure
Fast acting intravenous vasodilators are often given to postoperative hypertensive patients in the hospital. Due to varied patient response to different concentrations of a drug, vasodilators must be titrated during infusion. Currently, vasodilator titration is adjusted manually over time: a clinician periodically checks patient blood pressure and adjusts the infusion rate of vasodilator from a smart infusion pump based on hospital protocols written from package inserts provided by the drug manufacturer. This method of titration is time consuming for the clinician and can be inaccurate and imprecise, which clearly presents an opportunity for automating the drug infusion rate adjustment.
This is a problem that has been investigated by many different researchers in the past. Sodium nitroprusside was used as a drug for decades for this type of problem and it has been well characterized. A model for the patient’s response to the administration of the drug has been developed by Slate et al.(Slate, & Sheppard, 1982; Slate, Sheppard, Rideout, & Blackstone, 1980) and has been used by many researchers for controller design (Isaka, & Sebald, 1993). Close to one hundred publications of different control strategies using sodium nitroprusside had been reported by 1993 (Isaka, & Sebald) and papers are still published on this topic today. However, sodium nitroprusside is becoming less clinically relevant and is being replaced by Cleviprex (Grossman & Messerli, 2011) as the drug of choice for treating postoperative hypertension. The effect that Cleviprex has on the dynamic behavior of the mean arterial blood pressure of a patient differs significantly from what is known for sodium nitroprusside, and existing control strategies may not be appropriate when Cleviprex is used.
This work develops a dynamic model of the mean arterial blood pressure of patients who are given Cleviprex and then uses this model to develop a controller. It should be noted that this work is the first that develops a model for Cleviprex for treating postoperative hypertension. Data for model development is taken from Bailey (2002) and Singla (2008). Cleviprex is a fast acting drug and the response of a patient to infusion of the drug can be appropriately captured by a first order transfer function. However, there is significant patient-to-patient variability and the gain of the model can vary by an order of magnitude. This is a point that needs to be addressed for controller design: the patient is first given a continuous low dose of the drug and the response is analyzed after 3 minutes. The patient is then categorized as sensitive, nominal, or insensitive in his/her response to the drug and a separate PI controller has been designed for each case. It is important to note that this model identification step is possible due to the fast acting nature of Cleviprex, whereas this was impractical for sodium nitroprusside. In a next step, the chosen PI controller is used within a closed-loop control framework to lower the blood pressure to a desired level. Specific attention has been paid to include clinically-relevant design parameters such as upper/lower bounds for desirable blood pressure, upper limits on the drug dosage, keeping the infusion rate constant once the blood pressure has stabilized, and most importantly, alarming the clinician to unexpected behavior.
The controller is tested with the developed patient model using Monte Carlo simulations. Patient sensitivity, weight, and target blood pressure range are varied during testing. The controller proves effective for reducing the blood pressure for 98% of simulated patients while not exhibiting adverse effects, e.g., extensive overshoot, in any of the investigated cases.
Bailey, J. M., Lu, W., Levy, J. H., Ramsay, J. G., Shore-Lesserson, L., et al., “Clevidipine in Adult Cardiac Surgical Patients,” Anesthesiology,96 (5), pp. 1086-1094 (2002).
Grossman, E. and Messerli F. H., “Hypertensive Emergencies,” Chapter 4.6, in “Oxford Desk Reference: Cardiology,” H. F. Tse, G. Y. Lip, and A. J. S. Coats, Eds., University Press, Oxford, NY, pp. 137-138 (2011).
Isaka, S., and Sebald, A. V., “Control Strategies for Arterial Blood Pressure Regulation,” IEEE Transactions on Biomedical Engineering,40 (4), pp. 353-363 (1993).
Singla N., Warltier, D. C., Gandhi, S. D., Lumb, P. D., Sladen R. N., et al., “Treatment of Acute Postoperative Hypertension in Cardiac Surgery Patients: An Efficacy Study of Clevidipine Assessing Its Postoperative Antihypertensive Effect in Cardiac Surgery-2 (ESACPE-2), a Randomized, Double-Blind, Placebo-Controlled Trial,” Anesthesia & Analgesia,107 (1), pp. 59-67 (2008).
Slate, J. B., and Sheppard, L. C., “Automatic Control of Blood Pressure by Drug Infusion,” IEE Proceedings-A-Science Measurement and Technology,129 (9), pp. 639-645 (1982).
Slate, J. B., et al., “Closed-loop Nitroprusside Infusion: Modeling and
Control Theory for Clinical Application,” IEEE International Symposium on Circuits and Systems 1980, pp. 482-488 (1980).