Study on diagnosis criteria of fire-heat syndrome based on receiver operating characteristic curve and principal component analysis

Abstract

Objective

To establish the diagnostic quantitative criteria for fire-heat syndrome (FHS) of Chinese medicine (CM) based on the receiver operating characteristic (ROC) curve and principal component analysis (PCA).

Methods

The symptoms and signs of FHS cases and healthy subjects from Guangzhou, Henan and Hunan of China were collected through questionnaire, and the diagnostic quantitative score tables were established for the three regions, respectively, with the method of maximum likelihood analysis. The homogeneity test was then performed on the diagnostic score tables for the three regions with ROC curve, and the diagnostic efficiency of diagnostic score tables for the three regions was compared with the prospective test and retrospective test. The method of PCA was adopted to obtain the analysis matrix for classifying the tapes of FHS.

Results

Twenty-seven elements of FHS were confirmed through Chi-square test, and the diagnostic score tables for the three regions were established with the method of maximum likelihood analysis on the basis of the collected case data. According to the ROC curve test, the areas under ROC curve of Guangzhou diagnostic score table assessment with candidates in Guangzhou, Henan and Hunan were 0.998, 0.961 and 0.956, respectively. It showed that the diagnostic efficiency of Guangzhou diagnostic score tables was the highest one. With the prospective test, the area under ROC of Guangzhou diagnostic score table was 0.949, and more than any other diagnostic score table. By PCA, FHS was classified into excess fire and deficiency fire, and then classified into syndrome of flaring up of Heart (Xin) fire, syndrome of Lung (Fei)-Stomach (Wei) excess fire, syndrome of deficiency of Liver (Gan)-yin and Kidney (Shen)-yin, and syndrome of deficiency of Lung-yin from the view of viscera. In the retrospective test, the consistency with clinicians' diagnosis was 69.4%, and in the prospective test, it was 70.1%.

Conclusions

The Guangzhou diagnostic score table could be used as the recommended criteria for the diagnosis of FHS. The classification of FHS was basically in conformity with the clinical situation.