Detecting Adverse Drug Interactions with Multivalued Dimension Cube Technology

• Main Authors: Jhih-Wei Du, 杜治緯
• Other Authors: Wen-Yang Lin
• Format: Others
• Language: en_US
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/35748749552646527803

碩士 === 國立高雄大學 === 資訊工程學系碩士班 === 100 === Adverse Drug Reactions (ADRs) are uncomfortable or harmful side effects yielded by normal drug doses of usage. Indeed, some serious reactions may even lead to death. Many countries thus have set up ADR reporting systems to collect as possible all ADR events. As time passes, the number of reports grows dramatically, making manual analysis of these data impossible. Although in the past years, many statistical or data mining approaches have been proposed to detect suspected ADRs, most of them are very time-consuming and/or unable to detect ADRs caused by drug interactions. In this thesis, we propose the concept of multivalued dimension contingency (MDC) Cube to facilitate multidimensional, fast and online detection of adverse drug interactions. Experiments conducted on the FDA AERS data set show that our MDC cube-based method is significantly faster than the state-of-the-art ABCM-MS method.