Recurrence impact of primary site and pathologic stage in patients diagnosed with colorectal cancer

• Main Authors: Wen Chien Ting, Yen Chiao Angel Lu, Chi Jie Lu, Chalong Cheewakriangkrai, Chi Chang Chang
• Format: Journal
• Published: 2018
• Subjects: Decision Sciences; Engineering
• Online Access: https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85051539067&origin=inward; http://cmuir.cmu.ac.th/jspui/handle/6653943832/58604
• Institution: Chiang Mai University

© 2018, Chinese Society for Quality. All rights reserved. Detection of cancer recurrence for events of asymptomatic is highly related to the survival. In this study, we considered the variable screening mechanisms and four data mining techniques. The pathological data were obtained from Cancer Center of Chung Shan Medical University Hospital. Results show that primary site and pathologic stage are important independent risk factors. Before variable screenings showed that the highest of average accuracy and area under the curve (AUC) were: C5.0. Screening results of the colon site, the accuracy of < IIb stage was the highest with support vector machine (SVM) (0.91), and that of ≥ IIb stage was the highest with extreme learning machine (ELM) (0.86). In the rectum site, the accuracy of < IIb stage was the best with ELM (0.96), and that of ≥ IIb stage was the highest with multivariate adaptive regression splines (MARS) (0.89) and ELM (0.89). The results of this study provide that for recurrence detection in colorectal cancer patients can be used by clinicians to recommend adjuvant treatment.