Abstract
Objective:To construct a predictive maintenance model based on real-time monitoring data for medical equipment,so as to enhance precision and accuracy of managing medical equipment,and improve quality of clinical services.Methods:Data of operation,environment,faults and usage of equipment,and the data of diagnosis and treatment of patients were collected.A predictive maintenance model for medical equipment was constructed through conducting potential fault identification based on One-class SVM algorithm.A hierarchical maintenance strategy was formulated according to different levels of risk,so as to manage these equipment.The used 2,260 medical equipment in clinical work at Northern Theater Command of People's Liberation Army of China from 2021 to 2024 were selected.The conventional management mode was adopted to manage 2,035 equipment during 2021 and 2022,and the management mode with predictive maintenance model(predictive maintenance management mode)for medical equipment was adopted to manage 2078 equipment(included 1853 equipment that were managed by conventional management mode before 2023,and added 225 equipment after 2022)during 2023 and 2024.The operational efficiency of equipment,benefits of maintenance management,and quality of clinical service of the two management modes were compared.A self-made satisfaction questionnaire was used to investigate the satisfaction of users for clinical service of equipment.Results:The duration of normal operation of equipment that were managed by using the predictive maintenance management mode was(55.13±19.32)d,which was longer than that of the conventional maintenance management mode,and the difference was significant(t=3.361,P<0.05).The frequency of occurring fault,average downtime of equipment,and frequency of detecting abnormality of key parameter of predictive maintenance management mode were respectively(0.52±0.27)times·month-1·unit-1,(2.72±0.80)%and(1.21±0.72)times·month-1·item-1,all of which were lower than those of the conventional maintenance management mode,and the differences were statistically significant(t=3.361,3.312,4.396,4.970,P<0.05).The decreased degree of average maintenance cost of equipment,and the frequency of inventory turnover of the predictive maintenance management mode were all higher than those of the conventional maintenance management mode,and the differences were significant(t=4.016,4.732,P<0.05).The time of maintenance and management,and the frequency of emergent repair of the predictive maintenance management mode were lower than those of the conventional maintenance management mode,and the differences were significant(t=5.121,4.137,P<0.05).The frequency of occurring safety events of equipment of the predictive maintenance management mode was lower than that of the conventional maintenance management mode,and the difference was significant(t=6.316,P<0.05).The average accuracy rate of early warning for fault of the predictive maintenance management mode was higher than that of the conventional maintenance management mode,and the difference was significant(t=7.043,P<0.05).The satisfaction score of users for clinical service of equipment of adopting the predictive maintenance management mode was higher than that of adopting the conventional maintenance management mode,and the difference was statistically significant(t=4.241,P<0.05).Conclusion:The application of the predictive maintenance management model based on real-time monitoring data for medical equipment can improve operational efficiency of equipment,and benefits of maintenance management,and enhance the quality of clinical service of equipment,and improve the satisfaction of users for using equipment.关键词
预测性维护/实时监测数据/时空加权/随机森林/故障预警Key words
Predictive maintenance/Real-time monitoring data/Spatiotemporal weighting/Random forest/Early warning for fault分类
医药卫生
