The present invention provides a time sequence data processing method based on dynamic network diagram analysis. The method comprises: analyzing metabonomics queue data from the perspective of a network; analyzing a correlation among variables; constructing a metabolic network according to a dynamic change of the correlation among the variables; and analyzing and determining warning information of occurrence of diseases (such as a malignant tumor) according to a dynamic concentration change and a change of a network topology structure. The time sequence data processing method based on the dynamic network diagram analysis eliminates a disadvantage that characteristic dynamic change information is ignored when a static analysis method is adopted to process time sequence data of metabonomics. Moreover, compared with an algorithm of focusing on finding a monomolecule marker, in the method provided by the present invention, the change of the correlation among the variables over time is considered, and key turns of diseases are analyzed and determined, thereby facilitating a study on pathogenic mechanisms of diseases, and laying a foundation for early diagnosis and a prognostic study of diseases.