Study on evaluation of the algorithms for Adverse event signal detection with drug interaction detective system

2015 Fiscal Year Final Research Report

• Project/Area Number: 25460845
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Medical and hospital managemen
• Research Institution: Kagoshima University
• Principal Investigator: Muranaga Fuminori 鹿児島大学, 医歯学域医学部・歯学部附属病院, 講師 (00325812)
• Co-Investigator(Kenkyū-buntansha): KUMAMORO Ichiro 鹿児島大学, 医歯学域医系, 教授 (40225230) ; UTO Yumiko 鹿児島大学, 医歯学域医系, 准教授 (50223582) ; IWAANAKUCHI Takashi 鹿児島大学, 医歯学域医学部・歯学部附属病院, 助教 (80619198)
• Project Period (FY): 2013-04-01 – 2016-03-31
• Keywords: データマイニング / 医薬品相互作用 / 医薬品有害事象 / 有害事象シグナル検知 / データウェアハウス

Outline of Final Research Achievements

[Background] In the postmarketing surveillance (phase 4) of the new medicine,detecting the signal of the adverse event at postmarketing surveillance early,is essential to minimize the victim by the harmful effect.[Objectives] We investigate data mining technique that is optimal for the signal detection of the adverse event of the drug interaction.[Methods]We analyze a real prescription record,and develop a medication history simulator, and made 100,000 medication history data.We examined the utility of the association analysis, the Bayesian network analysis, the neural network analysis.
[Result and discussion] By the association analysis,we narrowed it down to the record which an adverse event developed and analyzed it and succeeded. By the Bayesian network analysis,we failed. By the neural network analysis, we were able to discover a candidate of the assumption combined medicine.

Free Research Field

医療情報学