基于监护仪质控大数据的性能预测模型初探
|
摘要
目的:从离散、模糊的监护仪质控数据中找出规律和相关性,建立较准确的预测模型并实现预估监护仪性能变化的目的。方法:以监护仪的血氧饱和度数据为例,充分研究数据属性,优化数据结构,利用ARIMA(autoregressive integrated moving average)时间序列模型和SPSS统计工具确定模型参数和总体误差。结果:得到了最优模型拟合参数值和预估趋势曲线,符合实际数据特性。模型预测值与实际值的最大绝对误差百分比小于3%,能够较准确地预测血氧电路的性能。结论:实验表明建立的预测模型拟合度较高,可近似反映质量及性能趋势,能及时描述设备状况,对医疗设备的预防性维护具有指导意义。
Objective To find the law and correlation from the discrete and fuzzy quality control data, to establish an accurate model and to estimate the future performance changes of the monitor. Methods Taking SpO_2 data of monitor as an example,data attributes were fully studied, data structures were optimized, and the ARIMA model of time sequence and SPSS statistical tools were used to determine model parameters and overall errors. Results It's shown the optimal model fitting parameters and estimate trend curve were obtained, which conformed to the actual data characteristics. At the same time, the maximum absolute error percentage of the model prediction value and the actual value was less than 3%, which could be considered to be accurate for predicting the performance of SpO_2 circuits. Conclusion The experiment indicates that the prediction model has a high fitting, which can approximately reflect the trend of quality and performance, timely express the equipment status, and is of significance for PM of medical equipment.
Objective To find the law and correlation from the discrete and fuzzy quality control data, to establish an accurate model and to estimate the future performance changes of the monitor. Methods Taking SpO_2 data of monitor as an example,data attributes were fully studied, data structures were optimized, and the ARIMA model of time sequence and SPSS statistical tools were used to determine model parameters and overall errors. Results It's shown the optimal model fitting parameters and estimate trend curve were obtained, which conformed to the actual data characteristics. At the same time, the maximum absolute error percentage of the model prediction value and the actual value was less than 3%, which could be considered to be accurate for predicting the performance of SpO_2 circuits. Conclusion The experiment indicates that the prediction model has a high fitting, which can approximately reflect the trend of quality and performance, timely express the equipment status, and is of significance for PM of medical equipment.
引文
[1]张美英,何杰.时间序列预测模型研究综述[J].数学的实践与认识,2011,41(18):189-195.
[2]麦鸿坤,肖坚红,吴熙辰,等.基于R语言的负荷预测ARIMA模型并行化研究[J].电网技术,2015,39(11):3 216-3 220.
[3]樊娇,冯昊,牛东晓,等.基于小波分析和GM-ARIMA模型的月度售电量预测[J].华北电力大学学报(自然科学版),2015,42(4):101-105.
[4]夏菁,张华勋,林文,等. ARIMA模型在疟疾发病率预测中的应用[J].中国血吸虫病防治杂志,2016,28(2):135-140.
[5]杜懿,麻荣永.不同改进的ARIMA模型在水文时间序列预测中的应用[J].水力发电,2018,44(4):12-14.
[6]顾刘金.应用SPSS软件实现多重线性回归分析[J].预防医学,2018,30(3):323-324.
[7]安胜利,张燕虹,陈征.应用SPSS软件实现二分类重复测量资料的GEE及GLMMs分析[J].南方医科大学学报,2012,32(12):1 777-1 780.
[8]刘洋,陶庭婷.正态分布的随机变量独立与不相关问题研究[J].山东农业大学学报(自然科学版),2018,49(2):339-341.
[9]郑浩然,潘雨青,李世伟,等.基于季节性ARIMA模型的小区供水预测[J].计算机应用与软件,2018,35(1):118-122.
[10]薛芳静,黄灏,许碧云,等.时间序列数据中的干预分析模型及SAS实现[J].中国卫生统计,2017,34(3):509-514.
[11]毛敏,刘建.基于多元线性回归的区域物流需求组合预测方法[J].物流技术,2015,34(7):163-165.
[12]许开成,毕丽苹,陈梦成,等.基于SPSS回归分析的锂渣混凝土抗压强度预测模型[J].建筑科学与工程学报,2017,34(1):15-24.
[2]麦鸿坤,肖坚红,吴熙辰,等.基于R语言的负荷预测ARIMA模型并行化研究[J].电网技术,2015,39(11):3 216-3 220.
[3]樊娇,冯昊,牛东晓,等.基于小波分析和GM-ARIMA模型的月度售电量预测[J].华北电力大学学报(自然科学版),2015,42(4):101-105.
[4]夏菁,张华勋,林文,等. ARIMA模型在疟疾发病率预测中的应用[J].中国血吸虫病防治杂志,2016,28(2):135-140.
[5]杜懿,麻荣永.不同改进的ARIMA模型在水文时间序列预测中的应用[J].水力发电,2018,44(4):12-14.
[6]顾刘金.应用SPSS软件实现多重线性回归分析[J].预防医学,2018,30(3):323-324.
[7]安胜利,张燕虹,陈征.应用SPSS软件实现二分类重复测量资料的GEE及GLMMs分析[J].南方医科大学学报,2012,32(12):1 777-1 780.
[8]刘洋,陶庭婷.正态分布的随机变量独立与不相关问题研究[J].山东农业大学学报(自然科学版),2018,49(2):339-341.
[9]郑浩然,潘雨青,李世伟,等.基于季节性ARIMA模型的小区供水预测[J].计算机应用与软件,2018,35(1):118-122.
[10]薛芳静,黄灏,许碧云,等.时间序列数据中的干预分析模型及SAS实现[J].中国卫生统计,2017,34(3):509-514.
[11]毛敏,刘建.基于多元线性回归的区域物流需求组合预测方法[J].物流技术,2015,34(7):163-165.
[12]许开成,毕丽苹,陈梦成,等.基于SPSS回归分析的锂渣混凝土抗压强度预测模型[J].建筑科学与工程学报,2017,34(1):15-24.