Association Analysis Among Treatment Modalities and Comorbidity for Prostate Cancer

Abstract

Prostate cancer is a common cancer treated with multi-modality. The combinations of modalities are numerous and complex. Clinical practice guidelines and rules have already been proven in many studies. However, the hypotheses of these studies came from physicians' and experts' experiences and observation. Association analysis, as an importance component of data mining, has been proved to be helpful for us to discover rules from big medical databases. We believe association analysis is able to help us to discover new rules between comorbidities and modalities in subjects of prostate cancer, so that employed it to analyze prostate cancer dataset derived from million people file of NHIRD. We successfully found six rules and rule 1,2,3,5,6 could be well explained with known knowledge and literatures, which were "Young prostate cancer patient who were spared from definite treatment tend to be spared from HT.", "TRUS is associated with younger age group, while TURP is associated with older Age.", "RT is associated with HT.", "CT is highly associated with RT.", "Hemiplegia, cerebrovascular disease, moderate to severe renal disease, diabetes with end organ damage is associated with TURP. Patients with TURP are associated with more comorbidity." We also discovered rule 4: "Younger patients who received HT is highly associated with previous RP.", which are still hypothesis and deserve our validation.