TITLE: FUZZY LOGIC MODEL FOR ESTIMATING THE DEPTH OF ANESTHESIA IN THE DOG

AUTHORS: Rob J. Roy, MD, DEng Sc¹, Xu-Sheng Zhang, PhD²
AFFILIATION: ¹Albany Medical College, Albany, NY; ²Rensselaer Polytechnic Institute, Troy, NY.

INTRODUCTION: We present a fuzzy knowledge model for quantitatively estimating the depth of anesthesia (DOA) and validate it by 30 experiments using 15 dogs with three different anesthetic regimens.
METHODS: We used nonlinear analysis to extract complexity measure C(n) [1] and approximate entropy ApEn [2], from the raw EEG signals and merged them together with the spectral entropy SE [3]. These measures form an input feature vector for training an Adaptive Network based Fuzzy Inference System (ANFIS)[4] to obtain fuzzy rules to express the relationship between these three derived parameters and the DOA. The performance of the model, its accuracy for one specific regimen, the variability in estimating DOA under different regimens, and the generalization ability across different regimens, was evaluated by conducting 30 experiments using 15 dogs with three different anesthetic regimens (propofol, isoflurane, and halothane).
RESULTS: The model demonstrated good performance in discriminating awake and asleep states (accuracy 90.3% for propofol, 92.7% for isoflurane, and 89.1% for halothane), and good generalization ability (85.9% across the three regimens).
DISCUSSION: Although we only grade the depth of anesthesia to awake (0.0) and asleep (1.0), after training the model can automatically estimate the intermediate states (between asleep and awake) and give a value between 1.0 and 0.0 to track the gradual transitions. Our studies indicate that the EEG contains sufficient information such that it will be feasible to obtain a reliable, continuous prediction of movement during surgical procedures by means of a method that is suited to the nature of the EEG signal.
ACKNOWLEDGEMENT: National Science Foundation under Grant BES-9522639 and by the Whitaker Foundation.
REFERENCES:

[1] IEEE Trans. on Inf. Theory, IT-22:75-81, 1976.

[2] J.Clin. Monit., 7:335-345, 1991.

[3] IEEE Trans. Biomed. Eng., 45(9):1186-1191, 1998.

[4] IEEE Trans. On Systems, man, and cybernetics, 23(3):665- 684, 1993.