The Application of Data Mining Techniques to Oral Cancer Prognosis

• Main Authors: Wan-ting Tseng, 曾婉婷
• Other Authors: Shyun-Yeu Liu
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/96009429245830375340

碩士 === 國立中正大學 === 資訊管理所暨醫療資訊管理所 === 98 === In recent years, the problem of oral cancer in Taiwan has soared to a new high. During the last five years, there has been a 30% increase in the incidence of oral cancer that translated to a 25% rise in mortality rate. In addition, the annual Taiwan male oral cancer incidence was ranked third in the world. Also, in the newest cancer report, the mortality rate of oral cancer has been the sixth major cancer type in cancer death. Moreover, oral cancer mortality is ranked fourth in cancer death of male patients. Due to the increase of oral cancer in Taiwan it is now considered an important public health topic. The survival rate and chance of treatment of oral cancer patient is affected by the late discovery, time factor and several other common factors. In addition, there are several treatment procedures that may influence the patient''s prognostic situation, recurrence and shift in the patient’s treatment in different situations and may bring in different medical cost. No matter if it is clinician or patient, the survival of oral cancer patient and their therapeutic success or failure are related with medical quality. Some information can be obtained from the patient and tumor organization, such as learning the infringement level of the tumor, and the model to treat it are great topics in the clinic. So, if we can understand the prediction factor of oral cancer it increases the patient''s survival rate and can expand medical treatment in the future. This could then be a reference material in health policy making, clinical management and the depth and broadness of cancer prevention and cure for oral cancer. Information technology has been widely utilized as a valuable tool to produce diagnosis guides and teaching aids in clinical studies. As data mining techniques becoming mature, their applications to medical field offer decision support to medical personnel in areas such as diagnosis, treatment, and care of patients. Two data mining tools, namely decision tree and artificial neural network, were used to analyze the historical cases of oral cancer, and their performance was compared with that of logistic regression, the popular statistical analysis tool. Both decision tree and artificial neural network models showed superiority to the traditional statistical model. However, as to clinician, the trees created by the decision tree models are relatively easier to interpret compared to that of the artificial neural network models. Cluster analysis also discovers that those stage 4 patients whose also possess the following four characteristics are having an extremely low survival rate: pN is N2b, level of RLNM is level I-III, AJCC-T is T4, and cells mutate situation (G) is moderate.