The Study And Implementation Of Smart Monitoring And Diagnosing Methods For The Complications In Anesthesia

During a routine surgery an anesthetist is in charge of maintaining patient's physiological state in sound condition through vigilant analysis of multiple channels of patient data. Anesthetists working in the operating theatre are responsible for carrying out a multitude of tasks which requires constant vigilance and are often overloaded with data from the monitoring equipment, thus lead to anesthesia related mishaps. Human errors contribute to a large portion of the anesthesia related mishaps and they can be easily prevented by providing decision support to the anesthetists. This paper firstly presented two algorithms for the detection and diagnosis of absolute hypovolemia which may occur in anesthesia, and then the design and implementation of the monitoring and diagnosing system based on virtue instrument was introduced. In addition, another noninvasive detecting algorithm for vocal cord paralysis which is a complication of anesthesia was also discussed in detail.A new algorithm named probabilistic alarm algorithm was developed for the diagnosis of absolute hypovolemia. This algorithm was a real time, statistic-based algorithm and it processed and analyzed the patients'data, such as blood pressure, heart rate, pulse volume and expired carbon dioxide etc. The adverse changes in these physiological variables detected by the probabilistic alarm algorithm are expressed in multiples of standard deviations (SD) and the sensitivity of the probabilistic alarms can be tuned by specifying appropriate alarm limits. Generating a probabilistic alarm from sequential physiological measurements is mathematically straightforward and is a rapid statistical tool for detecting statistically significant physiological variations. It was anticipated that the computational simplicity of the probabilistic algorithm and the transparency of the reasoning behind the probabilistic alarm results would give the anaesthetist more confidence.Another computer aided diagnosis algorithm using fuzzy logic was discussed. In this algorithm, a technology called fuzzy matching course was developed to discriminate absolute hypovolemia with other complications with similar symptoms. The accuracy and applicability of the algorithm were enhanced through the fuzzy matching course by incorporating the trend and range of patients'data.SPV represents the change in the peak blood pressure (systolic BP) values during a respiratory cycle and it could be a fair assessment of the cardiovascular status for the patients on mechanical ventilation. The SPV algorithm was researched and it was one of the two redundant diagnosing modules of the developed system.For the early detection of vocal cord paralysis in anesthesia, a noninvasive diagnosing method using wavelet packet analysis (WPA) and least squares support vector machines (LSSVM) was developed. The proposed method was compared to the hoarseness diagram method, which was reported as an objective voice quality evaluation approach and can be used for pathological voice discrimination. During the research, an improved cross validation algorithm was discussed for the LSSVM parameter selection and it can effectively decrease the computation complexity of LSSVM. Experiments showed that the proposed method had a higher accuracy, sensitivity and specificity than HDm and it could be a clinical screening tool to detect damage to vocal structures that arises during surgery.Based on the analyses and implementations of the proposed diagnosing algorithms, a smart monitoring and diagnosing system (SMDS) in anesthesia was designed with virtue instrument technology. Through real time testing and Kappa analysis, the designed system was proved to have a substantial level of agreement with the anesthetist's diagnoses, so it is of great value for reducing anesthetist's work load and improving the anesthesia safety.This study was supported by China National Natural Science Funds and the Coorperative Project of Auckland University of Technology (New Zealand). The designed SMDS system has been tested and used in the clinical environment at the Auckland City Hospital (New Zealand) and the Middlemore Hospital (New Zealand)..