摘要
目的:基于粒子群优化(PSO)算法构建设备风险管理模型,探讨其在急诊科心电监护设备管理中的应用价值.方法:采用PSO算法优化神经网络模型构建心电监护设备风险管理模型,收集北京市普仁医院心电监护设备运行风险数据进行归一化处理,并将2021年11月至2023年10月北京市普仁医院急诊科在用的30台心电监护设备,按照设备管理模式不同对其分别采用反向传播(BP)神经网络模型(简称传统BP模式,设备运行时段为2021年11月至2022年10月)和PSO算法的设备风险管理模型(简称PSO算法模式,设备运行时段为2022年11月至2023年10月)进行管理,比较两种管理模型设备风险故障识别效果(测试集与训练集)、警报风险控制效果和设备故障维修诊断用时.结果:采用PSO算法的测试集风险故障数据识别受试者工作特征(ROC)曲线下面积(AUC)值、准确率、灵敏度和特异度分别为0.869、93.6%、92.8%和95.1%,训练集风险故障数据识别AUC值、准确率、灵敏度和特异度分别为0.839、95.6%、97.9%和96.7%,均高于传统BP模式,差异有统计学意义(x2测试=3.691、4.023、3.557、3.409,x2训练=6.884、5.962、5.334、3.215;P＜0.05).采用PSO算法的心电监护设备警报阈值合格率和设备维护平均合格率分别为(98.61±3.07)%和(98.79±3.11)%,均高于传统BP模式,警报静音率为(1.14±0.27)%,低于传统BP模式,差异均有统计学意义(Z=11.831、10.020、21.141,P＜0.05).采用PSO算法的心电监护设备内部报修用时、外部报修用时、故障诊断用时和报修总用时分别为(1.21±0.96)、(3.18±1.09)、(5.08±1.93)和(10.95±2.81)min,均少于传统BP模式,差异有统计学意义(t=15.404、19.020、16.694、25.511,P＜0.05).结论:基于PSO算法构建的心电监护设备风险管理模型应用,能够提高心电监护设备风险故障数据识别灵敏度、特异度和准确性,提升警报阈值合格率和设备维护合格率,降低警报静音率,缩短故障诊断报修用时.
Abstract
Objective:To construct equipment risk management model based on particle swarm optimization(PSO)algorithm,and to discuss its application value in the management of ECG monitoring equipment in emergency department.Methods:The operation risk data of ECG monitoring equipment in the hospital were collected and normalized,and the PSO algorithm was used to optimize the neural network model to construct the risk management model of ECG monitoring equipment.30 ECG monitoring equipment in clinical use in the emergency department of Beijing Puren Hospital from November 2021 to October 2023 were selected and according to different equipment management modes,the backpropagation(BP)neural network model(referred to as the conventional BP model)and the PSO algorithm equipment risk management model(referred to as the PSO algorithm model)were used to manage the equipment respectively.The equipment risk fault identification effect,alarm risk control effect and equipment fault maintenance diagnosis time were compared between two management models.Results:The area under the curve(AUC)of the receiver operating characteristic(ROC)value,accuracy,sensitivity,and specificity of risk fault data identification in the test set using the PSO algorithm were 0.869,93.6%,92.8 and 95.1%,respectively,the AUC value,accuracy,sensitivity,and specificity of risk fault data identification in the training set were 0.839,95.6%,97.9%and 96.7%,respectively,which were higher than those of the conventional BP model,and the difference was statistically significant(x2test=3.691,4.023,3.557,3.409,x2training=6.884,5.962,5.334,3.215,P<0.05).The pass rate of ECG monitoring equipment alarm threshold and the average pass rate of equipment maintenance using PSO algorithm were(98.61±3.07)%and(98.79±3.11)%,respectively,which were higher than those of the conventional BP mode,and the alarm mute rate was(1.14±0.27)%,which was lower than that of the conventional BP mode,and the differences were statistically significant(Z=11.831,10.020,21.141,P<0.05).The internal repair time,external repair time,fault diagnosis time and total repair time of ECG monitoring equipment using PSO algorithm were(1.21±0.96)min,(3.18±1.09)min,(5.08±1.93)min and(10.95±2.81)min,respectively,which were all less than those of the conventional BP mode,the difference was statistically significant(t=15.404,19.020,16.694,25.511,P<0.05).Conclusion:The application of the risk management model of ECG monitoring equipment based on PSO algorithm can improve the sensitivity,specificity and accuracy of risk fault data identification of ECG monitoring equipment,improve the qualified rate of alarm threshold and equipment maintenance,reduce the silence rate of alarm,and shorten the time of fault diagnosis and repair.
国家科技期刊平台
NSTL
万方数据