Caminal P, Domingo L, Giraldo B F, Vallverdú M, Benito S, Vázquez G, Kaplan D
Biomedical Engineering Research Centre, Departament ESAII, Technical University of Catalonia, Spain.
Med Biol Eng Comput. 2004 Jan;42(1):86-91. doi: 10.1007/BF02351015.
This work proposed and studied a method of automatically classifying respiratory volume signals as high or low variability by means of non-linear analysis of the respiratory volume. The analysis used volume signals generated by the respiratory system to construct a model of its dynamics and to estimate the quality of the predictions made with the model. Different methods of prediction evaluation, prediction horizons and embedding dimensions were also analysed. Assessment of the method was made using a database that contained 40 respiratory volume signals classified using clinical criteria into two classes: low or high variability. The results obtained using the method of surrogate data provided evidence of non-linear determinism in the respiratory volume signals. A discriminant analysis carried out using non-linear prediction variables classified the respiratory volume signals with an accuracy of 95%.
这项工作提出并研究了一种通过对呼吸量进行非线性分析来自动将呼吸量信号分类为高变异性或低变异性的方法。该分析使用呼吸系统产生的呼吸量信号来构建其动力学模型，并估计该模型所做预测的质量。还分析了不同的预测评估方法、预测范围和嵌入维度。使用一个数据库对该方法进行评估，该数据库包含40个根据临床标准分类为低变异性或高变异性两类的呼吸量信号。使用替代数据方法获得的结果证明了呼吸量信号中存在非线性确定性。使用非线性预测变量进行的判别分析对呼吸量信号进行分类的准确率为95%。
