基于设备特性的全自动血球分析仪质量管理研究  

Research on quality management technology of fully automatic blood cell analyzer based on device characteristics

作　　者：孔琼 姜艳[1] 唐玲[1] 李辉[1] Kong Qiong;Jiang Yan;Tang Ling;Li Hui(Medical Laboratory Center,The First Affiliated Hospital of Xinjiang Medical University,Urumqi 830000,China)

机构地区：[1]新疆医科大学第一附属医院医学检验中心,乌鲁木齐830000

出　　处：《中国医学装备》2024年第10期123-128,共6页China Medical Equipment

摘　　要：目的:构建基于设备特性的全自动血球分析仪质量预测诊断模型,探讨其在全自动血球分析仪动态质量管理控制中的应用价值。方法:基于设备特性采用长短期记忆(LSTM)模型和Softmax分类器,运用Adam优化算法构建全自动血球分析仪质量预测诊断模型(简称质量预测诊断模型),设计18个设备特征参数作为输入变量,对设备常见的10种质量问题进行预测诊断。选取2021—2022年新疆医科大学第一附属医院检验中心在用的5台全自动血球分析仪数据为样本,进行模型训练和测试,评估模型预测正确率和诊断精确度,并用模型输出结果观察2023年1—6月5台全自动血球分析仪设备质量管理辅助效果。采用正确率和精确度评价质量预测诊断模型性能,对比模型辅助全自动血球分析仪质量管理前后周平均故障数降幅。结果:质量预测诊断模型应用于5台设备预测诊断,预测正确率分别为97.5%、95.9%、96.3%、95.3%和95.2%,5台设备预测诊断正确率比较差异无统计学意义(P>0.05);5台设备预测诊断精确度分别为92.6%、89.7%、91.1%、92.4%和91.1%,5台设备预测诊断精确度比较差异无统计学意义(P>0.05)。应用模型辅助5台设备质量管理控制后,周平均故障数降幅分别为67.35%、68.36%、69.72%、68.97%和67.47%。结论:基于设备特性的全自动血球分析仪质量预测诊断模型可根据设备特征值准确预测设备质量问题,正确诊断引发质量问题的原因并提出相应的处理措施,应用于设备质量管理控制可有效降低故障发生率,并适用于不同设备。Objective:To construct a quality prediction and diagnosis model for fully automatic blood cell analyzers based on equipment characteristics,and to explore its application value in dynamic quality management and control of fully automatic blood cell analyzers.Methods:Based on the characteristics of the equipment,the quality prediction and diagnosis model for automatic blood cell analyzer was constructed by using the Long Short Term Memory(LSTM)model and Softmax classifier and Adam optimization algorithm,and 18 equipment characteristics parameters were designed as input variables to predict and diagnose 10 common quality problems.The data of five automatic blood cell analyzers in clinical use in the Laboratory Center of the First Affiliated Hospital of Xinjiang Medical University from 2021 to 2022 were selected as samples for model training and testing,and the prediction accuracy and diagnostic accuracy of the model were evaluated,and the auxiliary effect of quality management of five automatic blood cell analyzers from January to June 2023 was observed with the model output results.The performance of the quality prediction and diagnosis model was evaluated by accuracy and precision,and the reduction in the average number of weekly failures before and after the model-assisted quality management of the fully automatic blood cell analyzer was compared.Results:The model was applied to the prediction diagnosis of five devices,and the prediction accuracy rates were 97.5%、95.9%、96.3%、95.3%and 95.2%,respectively,and there was no significant difference in the prediction and diagnosis accuracy of the five devices(P>0.05).The diagnostic accuracy was 92.6%、89.7%、91.1%、92.4%,and 91.1%,respectively,with no statistically significant difference(P>0.05).The predictive diagnostic accuracy of the five devices was 92.6%,89.7%,91.1%,92.4%and 91.1%,respectively,and there was no significant difference in the predictive diagnostic accuracy of the five devices(P>0.05).After applying the model to assist the quality management c

关 键 词：设备特性 全自动血球分析仪 质量管理 人工智能 

分 类 号：R197.39[医药卫生—卫生事业管理]