ABSTRACT
Many clinical studies showed that the traditional Chinese medicine (TCM) syndromes in stroke have been dynamically changing since the onset of the disease. The changing of TCM syndromes can be attributed to multiple correlative factors such as age, sex, area distribution, underlying diseases, and constitutional factor. Data-driven methods involving multivariate statistical methods and descriptive approach have been used to analyze the regularity of dynamically changed TCM syndromes of stroke. However, expressing non-linear relationship between symptom or correlative factors and syndrome patterns by data-driven models is challenging. Model-driven methods involving artificial neural networks and Bayesian networks are new methods for studying the changes in TCM syndromes in patients with stroke. In this review, the authors summarized the studies of dynamically changed patterns of stroke syndromes based on data-driven methods and some clinical trials on TCM syndromes based on model-driven methods. Further studies are needed to improve the understanding of the dynamically changing regularity of TCM syndromes for stroke by using model-driven methods so as to develop appropriate and timely TCM treatments.