论文部分内容阅读
“M+N”理论是针对光谱分析提出的用于提高预测精度的一种理论方法。将被测量、M种非测量组分和N种外界干扰一同归于整个测量系统中,系统全面地考虑“M”因素和“N”因素对测量精度的影响。建立基于“M+N”理论的血氧饱和度人体试验系统,采用动态光谱法消除个体差异及不同测量部位(“N”因素中的系统误差)对预测精度的影响;利用差值提取法消除运动伪差(“M”因素中的随机误差),获得了62组有效样本数据。采取偏最小二乘法建模预测,预测值与真值的相关系数为0.796 8,相对误差为±0.026 6;而目前临床上普遍采用的血氧监护仪与真值的相关系数为0.595 7,相对误差为±0.076 0。表明基于“M+N”理论的近红外血氧饱和度无创测量方法其预测精度明显优于血氧监护仪。
“M + N ” theory is a theoretical method proposed for spectral analysis to improve the prediction accuracy. Will be measured, M kinds of non-measured components and N kinds of outside interference together into the entire measurement system, the system fully consider the “M” factor and “N” factors on the measurement accuracy. The system of human body oxygen saturation test based on “M + N ” theory was established. The dynamic spectroscopy was used to eliminate the individual differences and the influence of different measurement sites (systematic errors among “N” factors) on the prediction accuracy. The value extraction method eliminates motion artifacts (random errors in “M ” factors) and obtains 62 sets of valid sample data. The partial least square method was used to predict the model prediction. The correlation coefficient between the predicted value and the true value was 0.796 8, the relative error was ± 0.026 6. At present, the correlation coefficient between the blood oxygen monitor and the true value commonly used clinically is 0.595 7, The error is ± 0.076 0. The results show that the prediction accuracy of NIRS based on “M + N ” theory is better than that of oximetry.