The influence of two different drug infusion profiles on the pharmacodynamics model performance

Abstract

To model the effect of anesthetic drugs on the Bispectral Index (BIS) of the EEG is of great importance for a reliable predictive response model during surgery. In this study, the impact of using two different drug infusion profiles in a pharmacodynamics interaction model was studied and the methods were compared with respect to their performance. Clinical data of 22 patients were considered. The interaction model was optimized per patient using nonlinear least squares during the induction of anesthesia, and tested for prediction abilities of test patients. In the optimization phase, all models could follow the BIS trend with errors not significantly differ- ent from zero. In the test data the choice of drug infusion pro- file proved to have a significant impact, showing that the per- formance is greatly influenced by the interaction. Results also show a time delay between the BIS signal and the Modeled BIS in both groups. This delay corresponds to a delay in the dy- namics of the patient and could be related to the delay in BIS processing time. This work is an important step to predict the effect of anesthetic drugs.