为了对冠心病进行快速筛查,运用近红外光谱仪,采集157名志愿者舌尖近红外反射光谱并进行反射率归一化,同时记录相对应的临床诊断信息.将样本分为校正集和预测集2部分.运用主成分分析结合神经网络法以及偏最小二乘2种方法建立分类预测模型.实验发现,2种方法预测准确率均达到100%.实验结果表明,舌体近红外反射光谱法运用于冠心病快速筛查是可行的.
In order to rapidly screen coronary heart disease (CHD), the reflection spectrum on the tongue tips of 157 volunteers was collected using near-infrared spectrometry and the normalized reflectivity was calculated. At the same time, the clinical information was recorded. Samples were divided into two groups, calibration set and prediction set. Two methods were used to set the classification model, namely, principle component analysis (PCA)combined with artificial neural net work (ANN) and partial least squares (PLS). Prediction accuracy of the two methods was 100%. Experiment results show that NIR reflection spectroscopy of tongue can provide a promising approach to the rapid screening of coronary heart disease.