RESUMO
The study on the effective core formulae (CEF) not only summarized traditional chinese medicine (TCM) treatment experience, but also helped reveal the underlying knowledge in the formulation of TCM prescriptions. The aim of the present paper was to investigate the method of data mining for the discovery of core effective formulae for lung cancer. In the present study, a prescription fingerprint approach was used to characterize the staged prescription information of patients. The D index was used to screen potential beneficial herbs. Then, based on a herbal compatibility network, the maximal clique searching algorithm (BK algorithm) and survival analysis were applied to discover CEF for lung cancer, and a mining analysis was made for the 322 cases from Longhua hospital. The correlation between prescriptions and survival time was analyzed by prescription fingerprints. Forty-three potentially beneficial herbs were obtained, and two CEFs were significant for the survival time by a parametric survival model based on lognormal distribution, the results were verified by a multivariate survival model. The rules of combination of the two CEFs basically conform to TCM onco-therapeutic theory of strengthening the body resistance and the actual conditions in clinic. All results showed that the established approach was feasible for discovering the core effective formulae for lung cancer and mining survival data for complex TCM onco-therapy.