Improvised Prediction And State Monitoring Of Anaesthesia Level Of Patients Using Deep-Learning Approach

Abstract

The anesthesiologist uses three different types of medicine to control the level of anesthesia during surgery: muscle relaxants, which are often used to suppress muscular reflexes, hypnotics to induce and maintain unconsciousness, and analgesics to block pain. These days, anesthesiologists may use instruments that measure unconsciousness in real time to determine the ideal dosage of hypnosis. These monitors typically use electrodes to attach to the patient's forehead and display a signal derived from the EEG activity of the patient. Based on the value of the signal, the anesthesiologist can assess the patient's state of unconsciousness. The amount of anesthesia provided depends directly on the patient's preoperative physical condition and intraoperative vital signs. It also has to consider the time the patient would need to recover after surgery. In order to safeguard their patients' vital signs during surgery, anesthesiologists must possess specialized knowledge and be quick thinkers, as each patient has distinct intraoperative conditions and symptoms. Thus, the proposed approach provides an effective means of predicting the patient's state of awareness during anesthesia. In this work, Ensemble Machine Learning techniques like SVM and Random Forest together with Deep Learning techniques like CNN and LSTM are used to determine the level of anesthesia and the accuracy found to be 96%.