Intelligent Model for Electromyogram (EMG) Signal Prediction During Anesthesia
José-Luis Casteleiro-Roca, Juan Albino Méndez Pérez, Andrés José Piñón-Pazos, José Luis Calvo-Rolle and Emilio Corchado
The use of engineering tools in anesthesiology is leading to the emergence of more efficient ways to administer drugs. Usually, their success greatly depends on the availability of reliable models to predict physiological variables response of patients. The aim of our research is to create a model to predict the evolution of the muscular relaxation measured through Electromyogram (EMG), monitored during the period of anesthesia under surgery. The patients EMG signal prediction is based on the Bispectral Index™(BIS™) and the propofol drug dose infusion. For achieving a good estimation of the EMG signal, a hybrid intelligent model was created by using clustering combined with regression techniques. It was employed a real dataset obtained from patients undergoing anesthesia during surgeries. The proposed model was validated and the obtained results were very satisfactory.
Keywords: EMG, BIS™, clustering, MLP, SVM