Mining Patient Profiles from Healthcare Social Media

• Main Authors: Meng-Hsiu Lee, 李孟修
• Other Authors: 李瑞庭
• Format: Others
• Language: en_US
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/91627748698652331240

碩士 === 國立臺灣大學 === 資訊管理學研究所 === 102 === With more and more user involvements in social network platforms, many kinds of social network applications have been rapidly growth. Healthcare social networks have become a popular type of social networks recently. They provide platforms for users to share their experience and for researchers to collect and analyze the data. In this thesis, we propose a framework to investigate the relationships among medical features (conditions, symptoms, treatments, effectiveness and side effects) for different types of patients in the healthcare social network. The proposed framework contains three phases. First, we extract the medical features of conditions, symptoms, treatments, effectiveness and side effects for each patient in the healthcare social network. Next, we modify the affinity propagation method (AP) to cluster patient profiles in a hierarchical manner, where we design a weighed scheme to increase the importance of less frequent but significant medical features. Finally, we mine frequent medical patterns for each cluster, and analyze the patterns mined and patient profiles in the resultant clusters. The experiment results show that the proposed framework can cluster similar patient profiles together, provide patients some quick references and help doctors to conduct in-depth analyses.