Abstract
Spray drying hot melt adhesive wall is mainly related to the characteristics of the material itself. The previous study of our group suggested that there was a nonlinear correlation between the physical properties and chemical compositions of Chinese Medical Herbs Aqueous (CMHDs) and the hot-melt adhesive wall of spray drying. In this study, we investigated 66 herbs from Chinese medicinal materials commonly used in Chinese pharmacopeia. Physical properties including dynamic viscosity, equilibrium surface tension, and dynamic surface tension (DST) were determined. Chemical compositions including the organic acids, low molecular weight saccharides, protein, total phenols, and tannins in their CMHDs were also analyzed. Then we conducted nonlinear-based data mining using a mutual information method based on entropy and association algorithm. Mutual information analysis showed that the score of DST10ms was higher than that of equilibrium surface tension, indicating that DST10ms was the physical property index that had the greatest influence on the hot melt adhesive wall of spray drying in this study. The higher the DST10ms, the higher the incidence of hot melt adhesive wall and the lower the yield of spray drying. The result of information entropy is shown the content of organic acids and low molecular weight saccharides among the chemical compositions directly affected the spray drying yield, which citric acid, malic acid, glucose, and fructose content had the greatest influence, and the results are in good agreement with those of previous experiments. The study also found that the water extraction liquid of fruit, flowers, seeds, and other parts of the medicine herbs has a low spray drying yield, the phenomenon of the observation is given priority to hot melt adhesive wall. The results of this study show that mutual information derived from association algorithms, based on information entropy, is useful to study correlations between parameters during the preparation of traditional Chinese medicines. Data mining has the potential to provide information for searching spray drying correlations by uncovering hidden rules between the physical and chemical properties of CMHDs, and the spray drying yield in a data set.