Extraction of best set of parameters derived from electroencephalogram for the prediction of awake and sleeps state of the patient

Abstract

The study describes a best set of parameters of electroencephalogram (EEG) for prediction of awake and anaesthetic sleep state of the patient during halothane anaesthesia. After obtaining the proper approval of hospital ethics committee and written consent from the patients, EEG data of 60 patients were recorded in normal awake and anesthetised sleep state. The recorded EEG data of all patients in each state was partitioned into different epochs of four-second duration. Algorithms were written to compute the 21 identified parameters of EEG, which are known to vary with variation in anaesthetic drug. Principal component analysis (PCA) is applied on these 21 parameters of EEG to find out the best parameters. Three parameters were identified which distinguish the awake and anesthetised sleep state of the patient. By using these identified parameters, whole data is divided into two different groups, i.e., training data and testing data. Artificial neural network (ANN) was trained with training data and target matrix. On testing with unknown epochs of testing data, this artificial neural network (ANN) has given 100% accuracy for the prediction of the state of the patient.