摘要Background: Drying is a necessary component of traditional Chinese medicine extracts. The heating principle of microwave vacuum drying is different from that of the conventional heat method. However, at present, there is paucity of information on the drying process of traditional Chinese medicine extract by microwave vacuum drying, and the results of such process are unclear. Methods:To study the dynamic changes in the chemical characteristics of microwave vacuum drying under different drying conditions, ultrahigh-performance liquid chromatography fingerprint profiles were established using Radix isatidis extract as a model drug and analyzed using similarity analysis, partial least squares-discriminant analysis, and semi-quantitative analysis. In addition, a backpropagation artificial neural network model was developed to predict the moisture ratio of the drying process. Results: Qualitative results showed that the similarity between different drying conditions was greater than 0.95, and 2 amino acid components (peaks 5 and 6) affected by process fluctuations were screened out. The quantitative results showed that the mass concentration of component 1 fluctuated after drying, while that of component 2 increased. The optimal backpropagation artificial neural network model structure used to predict the moisture ratio was 5-4-1, with regression and mean squared error values of 0.996 and 0.0003, respectively, after training, which were well fitted and had a strong approximation ability. Conclusion: Upon comparison of fingerprints and the evaluation of statistical methods, common components of Radix isatidis extract had little variation under different drying conditions, and the selected components provided a reference for the establishment of process evaluation indexes. The establishment of backpropagation artificial neural network provides a theoretical basis for the application of microwave vacuum drying technology and online monitoring of moisture ratio.