Real-time closed-loop control in a rodent model of medically induced coma using burst suppression

Abstract

Background: A medically induced coma is an anesthetic state of profound brain inactivation created to treat status epilepticus and to provide cerebral protection after traumatic brain injuries. The authors hypothesized that a closed-loop anesthetic delivery system could automatically and precisely control the electroencephalogram state of burst suppression and efficiently maintain a medically induced coma.

Methods: In six rats, the authors implemented a closed-loop anesthetic delivery system for propofol consisting of: a computer-controlled pump infusion, a two-compartment pharmacokinetics model defining propofol's electroencephalogram effects, the burst-suppression probability algorithm to compute in real time from the electroencephalogram the brain's burst-suppression state, an online parameter-estimation procedure and a proportional-integral controller. In the control experiment each rat was randomly assigned to one of the six burst-suppression probability target trajectories constructed by permuting the burst-suppression probability levels of 0.4, 0.65, and 0.9 with linear transitions between levels.

Results: In each animal the controller maintained approximately 60 min of tight, real-time control of burst suppression by tracking each burst-suppression probability target level for 15 min and two between-level transitions for 5-10 min. The posterior probability that the closed-loop anesthetic delivery system was reliable across all levels was 0.94 (95% CI, 0.77-1.00; n = 18) and that the system was accurate across all levels was 1.00 (95% CI, 0.84-1.00; n = 18).

Conclusion: The findings of this study establish the feasibility of using a closed-loop anesthetic delivery systems to achieve in real time reliable and accurate control of burst suppression in rodents and suggest a paradigm to precisely control medically induced coma in patients.

Fig. 1

Closed-loop anesthetic delivery system design for burst-suppression control. (A) Closed-loop anesthetic delivery system design. A burst-suppression probability (BSP) target is specified by the user (step 0) whereas an infusion pump maintains constant flow of propofol to the rodent through tail-vein intravenous catheter (step 1). Electroencephalogram (EEG) is recorded (step 2) and segmented into a binary time series representing bursts and suppressions (step 3). The binary time series is passed to the BSP filter to compute a real-time estimate of the BSP (step 4). The BSP estimate then feeds back (step 5) and is compared with the target BSP (step 6). The difference between the target and the current estimate, termed the error signal, is passed to the proportional-integral (PI) controller, which issues compensatory commands to the infusion pump (step 7). (B) Timeline of experiment. First, one or more boluses are administered to facilitate system identification (ID) of the rodent pharmacokinetics model parameters (15 min). After system identification, the system is switched to closed-loop operation and an initialization is undertaken to ensure that the system operates as expected (10 min). Finally, the BSP target tracing is initiated and control begins (60 min).

Fig. 2

Real-time segmentation of electroencephalogram (EEG) recordings. Segmenting clean (A) and noisy (B) EEG into bursts (black curves) and suppressions (gray curves). In each panel are: unprocessed EEG (1), filtered EEG (2), and binary time series (3). An amplitude threshold (red horizontal line) is applied to the filtered EEG to produce the binary time series. (A) Rat 6, showing a clean EEG with easily discernible, sharp bursts. (B) Rat 1, showing a noisier EEG with broader bursts.

Fig. 3

Electroencephalogram (EEG) segmentation and burst-suppression probability (BSP) estimation from a bolus infusion. (A) The unprocessed EEG. (B) Filtered EEG with threshold (red horizontal line). (C) Binary time series. (D) BSP filter estimate of the BSP time course. (E) Twenty-five second bolus of propofol that induced the EEG response in (A).

Fig. 4

System identification of pharmacokinetics model for rat 1. (A) Measured burst-suppression probability (BSP; gray curve) and fitted response (red curve). Gray line is the BSP estimated by the BSP filter and red line is the fit of the two-dimensional pharmacokinetics model, obtained by nonlinear least-squares fitting (see Materials and Methods). (B) Two bolus infusions, which induced the BSP responses in (A).

Fig. 5

Closed-loop control in six animals (rats 1–6 in A–F, respectively). Each upper subpanel shows burst-suppression probability (BSP) target trajectory (red line) and control BSP time course (black curve). Each lower subpanel shows instantaneous infusion rate. All permutations of the sequence 0.4, 0.65, and 0.9 are achieved. See table 1 for performance results.

Fig. 6

Assessment of closed-loop anesthetic delivery system reliability and accuracy using modified boxplot summaries of the absolute error and error distributions. (A) Modified boxplot summaries of the absolute error distributions at each level for the six animals. Whiskers are the 95th percentiles of the absolute error distributions. The lower (upper) border of the box is the 25th (75th) percentile and the middle line is the median. The closed-loop anesthetic delivery system is reliable (95th percentile <0.15) for all levels except for animal 2 at 0.65. (B) Modified boxplot summaries of the error distributions at each level for the six animals. Whiskers are the 2.5th and 97.5th percentiles of the error distributions. The closed-loop anesthetic delivery system was highly accurate (25th percentile < 0 < 95th percentile) for all levels except for animal 5 at level 0.4.