Home > Research


  • 2007-10-07 (日)

This research proposes a learning navigation system for anesthetists to apply interns to operate surgeries without errors and to study anesthetic practice in operation room. For this purpose, the system gives him or her anesthetic plan, warning, emergency, and alternative plan according to individual level. It has an engine that uses bayesian networks layer (BNL) model to represent anesthetic practice. BNL consists of 3 types of bayesian networks, which are activity bayesian networks, action bayesian networks, and operation bayesian networks, according to abstractive level of anesthetists. Using this system, anesthetic interns can keep "ready-to-hand action", which means that human continues anesthetic actions unconsciously and anesthetic adviser can know situation of his or her interns and situation of surgery outside the operation room.

Anesthetists overcome incidents and accidents with the system. First, anesthetic interns enter the plan into the system, such as the plan of operation of artificial joint replacement or cardiovascular surgery and so on. Second, the system interprets this plan and the interns receive warning depending on the plan of the surgery. For example, blood pressure usually changes in induction of anesthesia, so before this event, anesthetist receives the information of the incidents. Another example is that the system tells him to check the blood pressure when the surgeon uses the cement into knee joint. Third, if some unintentional emergency occurs, anesthetist can receive the emergency data, such as blood pressure will decrease or breathe stops for 2 minutes. Fourth, if the anesthetist fails to adhere to the plan in this emergency, the system gives him the alternative plan. For example, he receives the information that he should insert the ephedrine for the time being and so on. Or, if the anesthetist can not do anything, adviser comes to help him or her. Briefly, the system gives him the anesthetic plan, warning, emergency, and alternative plan in operation room at the same time when asking for the help of adviser.

The system is developed with BNL that gives the useful information depending on individual levels to anesthetists. BNL consists of three types of bayesian networks, which are activity bayesian networks to represent goals of anesthetists, action bayesian networks to represent anesthetic action in order to the goal, and operation bayesian networks to represent unconscious movement in the action. Using different levels of bayesian networks, anesthetists can get the appropriate information as to the individual levels and the situation of operation. Using the system with BNL, anesthetists can do the appropriate action with anesthetic plans, warning, emergency, and alternative plan and study the anesthetic practice in real situation, operation room.


TrackBack URL for this entry
Listed below are links to weblogs that reference
Research from narupeko.jp

Home > Research


Page Top