According to the idea of “correspondence of prescription and syndrome” in Treatise on Cold Damage，this study introduced the international concept of symptom cluster，which was relatively fixed and highly generalized，to objectively reflect syndrome.In addition，the types and characteristics of insomnia symptom clusters of effective and ineffective population were analyzed.This study is expected to provide methodologies for exploring symptom clusters of insomnia.Methods:Based on the real-world clinical data of insomnia and the widely accepted efficacy evaluation standard，the confounding factors were eliminated by propensity score matching and the effective and ineffective cases were screened.The text mining method was used to process information on the symptoms from effective and ineffective insomniacs and extract high-quality standardized symptom set.Then，the Lantern app was employed for latent tree analysis of the high-frequency symptoms.Results:Latent tree analysis classified insomnia symptoms into clusters and provided a mature model and algorithm for the objectivity and stability of symptoms clusters.In addition，different insomnia populations had different types and characteristics of symptom clusters，which laid a basis for the objectification of syndrome.Conclusion:Symptom clusters contain the main information of syndrome and the use of relatively fixed and highly generalized symptom clusters to supplement and replace “syndrome” for syndrome differentiation and treatment might be able to eliminate the instability and inconsistency among different doctors due to the difference in individual experience and academic school.This study is expected to provide methodology for related research.