Monitoring: Equipment and Computers
Introduction: The response to titanic stimulation can be predicted by a population derived interaction model (e.g. The Adaptive Neuro Fuzzy Inference System (ANFIS))1 or an individual measure derived from electroencephalogram (EEG) (e.g. qNOX, Quantium Medical, Barcelona, Spain) or a combination of both. This study evaluates which approach has the highest accuracy for predicting responsiveness.
Methods: The qNOX is derived from EEG matched with clinical signs from anesthetized patients. Four frequency ratios with optimal prediction probability of response to noxious stimuli were fed into an ANFIS Model, where the output was the qNOX.
We reused data from a previously published study2, including observations of response to titanic stimulation in 45 adult female patients, who were scheduled for gynecological surgery. Before stimulation they received a propofol effect-site concentration of 1.5 ug/ml in the three groups, while remifentanil was targeted to 0, 2 or 4 ng/ml respectively.
We calculated the prediction probability for qNOX, ANFIS model and qNOX + ANFIS combined model. The combination of the interaction model with qNOX was done using an ANFIS model as well.
Results: The results of the Pk analysis are shown in table 1.
Conclusion: The combination of qNOX and ANFIS had a significantly higher Pk =0.94 compared to respective performance of each method separately. Hence combining interaction surfaces with a direct measurement may be the optimal approach for accurate prediction of response to a noxious stimulus.
1. Struys MM, Vereecke H, Moerman A, et al Anesthesiology. 2003 Oct;99(4):802-12.
2. Gambús PL, Jensen EW, Jospin M, et al Anesth Analg. 2011 Feb;112(2):331-9.