|
|
October 16-20, 2010 San Diego, CA Home Abstract Archive Search Abstracts 2010 Session Grid 2010 Meeting Website Policy Statements ASA Website Feedback 
Visit Anesthesiology.org |
| Previous Abstract | Next Abstract |
| Printable Version |
|
A22
October 13, 2007 9:00 AM - 10:30 AM Room Room 302 |
| Dynamic Modeling of the Effect of Propofol and Remifentanil on BIS: The Influence of Heart Rate |
|
Catarina S. Nunes, Ph.D., Teresa Mendonca, Ph.D., David A. Ferreira, Ph.D., Luis Antunes, Ph.D., Amorim Pedro, M.D. Dep. Matematica Aplicada, Faculdade de Ciencias da Universidade do Porto, Porto, Portugal |
|
Background: The effect of drugs' interaction on the brain signal Bispectral Index (BIS) is of great importance for an anesthesia control drug infusion system. A previous study1 showed that baseline heart rate has influence on the amount of propofol required for loss of consciousness (LOC). In this study, the objective was to inspect the influence of patient's heart rate (HR) on the effect of the drugs on BIS. With this goal, the patient's heart rate was incorporated in a drug interaction model2,3. Methods: Data were collected in 45 neurosurgeries with propofol/remifentanil anesthesia, using RugLoopII® software every 5s from A2000XP®. The propofol effect-site concentration was predicted using PK Marsh/Diprifusor4 model, and the remifentanil concentration was predicted using Minto5. Anesthesia started with a constant infusion 200ml/h of propofol until LOC, thereafter propofol was changed according to BIS. The remifentanil infusion started at LOC. The BIS signal was prefiltered with a lowpass 2nd order Butherworth. The interaction model with the propofol potency dynamicaly adjusted to patient's heart rate (HR) was fitted to the data of each patient in the induction phase (first 15min), and parameters optimized using nonlinear least squares. The individual patient models were then used to predict BIS until recovery, using the drugs' concentrations. To identify groups of patients the fuzzy c-means clustering algorithm was applied to the obtained model parameters. Data as mean±SD. Results: The 45 patients were ASA 1/2, 51±16 years, 70±13 kg, 163±9 cm, 28 female, baseline HR 69±15 bpm. The individual models had a good performance in the induction (optimization) phase with statistical zero errors (P<0.05) in 40 patients. The average of absolute errors was 3.87±1.42, capturing the induction BIS trend in all patients (Fig. 1-B). When the individual models were used for prediction (from 15min to the end), performance, the average of absolute errors was 13.04±8.22. Fig. 1-C shows the model results for patient 26 and the HR signal, one can observe the HR influence on the BIS signal, wih positive correlation. Three clusters of model parameters were identified. Discussion: This model proved to be effective in modeling the induction BIS trend in all 45 patients, capturing unique characteristics (model parameters), e.g. patients that respond faster/slower to the same infusion dose. The results showed that the model with time changing parameters incorporating patient's heart rate has a better performance than a non adjusted model3. Three clusters of models were also identified using a clustering algorithm. These clusters will help to distinguish between different patients' dynamics. References: 1 – Anesthesiology 2006, 105: A620; 2- Anesthesiology 2003, 98:621-7; 3- Anesthesiology 2006, 105: A1201; 4- Br J Anaesth 1991 67:41-8; 5- Anesthesiology 1997 86:24-33. Acknowledgement: Portuguese Foundation for Science and Technology and UISPA-IDMEC.[figure1] Anesthesiology 2007; 107: A22 |
Figure 1
 |