| Most equipment-related anesthesia critical events, including the ones associated with the breathing circuit, could be prevented by checking the equipment before use. Standardized pre-use checklists have been developed and widely published. Even though most anesthesiologists are aware of them and their importance, many do not check out their equipment before using it. This dissertation discusses the motivation for, development of, and the testing of a new, advanced “smart” pre-use checkout system. It also describes an automated fault creation and data collection system used to acquire data for developing and testing of the pre-use checkout system.; The smart pre-use checkout was designed to detect and identify 22 anesthesia machine fault types and estimate their severity. Flow and pressure signals from three commonly used breathing circuit flow/pressure sensors are collected during a checkout sequence that takes less than 30 seconds to complete and requires minimal user-machine interaction or supervision. A hierarchically organized “generalists”/“specialist” artificial neural-network-based pattern recognition system detects and identifies the specific fault. It then evokes 1 of 10 algorithms to estimate the severity of the identified fault.; For the training of the artificial neural networks and development of fault quantification algorithms, a system that automatically creates faults in differing degrees of severity and collects data was designed and built. This system reduced the average time it took to collect data for one fault from 180 seconds for the manual system requiring constant supervision to 66 seconds, during which there was essentially no user interaction or supervision required.; Testing on 65,000 breaths covering almost 300 different combinations of faults, breathing circuits, and devices installed in the breathing circuit, the “smart” pre-use checkout system correctly identified 98.5% of the fault instances and had a low false positive rate of 1.7% of the “no fault” instances. In more than 960 of the cases the fault size was correctly classified.; The speed, efficiency, and ease-of-use of the “smart” pre-use checkout system, especially when compared to conventional procedures, could lead to a much better acceptance rate among anesthesia care providers than what is seen with conventional checkout methods. |
