Anomaly Detection on Patients Undergoing General Anesthesia

摘要

The importance of the infusion drug optimization in patients undergoing general anesthesia has led to the implementation of automatic control loops and models to predict the state of the patient. The appearance of any anomaly during the anesthetic process may lead, for instance, to incorrect drug administration. This could produce undesirable side effects that can affect the patient postoperative and also reduce the safety of the patient in the operating room. This study evaluates different one-class intelligent techniques to detect anomalies in patients undergoing general anesthesia. Due to the difficulty of obtaining data from anomaly situations, artificial outliers are generated to check the performance of each classifier. The final results give good performance in general terms.