Development of statistical method for case-cohort design when measurements of interests are known to be missing

2016 Fiscal Year Final Research Report

• Project/Area Number: 25330039
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Statistical science
• Research Institution: Shiga University (2016); Oita University (2013-2015)
• Principal Investigator: Izumi Shizue (大久保志津恵) 滋賀大学, データサイエンス教育研究センター, 教授 (70344413)
• Co-Investigator(Renkei-kenkyūsha): TAKEUCHI AYANO 慶應義塾大学, 医学部, 専任講師 (80511196) ; Cologne John 放射線影響研究所, 統計部, 研究員 (50344411)
• Research Collaborator: SUENAGA SATOSHI ; NAGATA DAIKI
• Project Period (FY): 2013-04-01 – 2017-03-31
• Keywords: 医薬生物 / ゲノム統計解析 / ビッグデータ活用 / ゲノム疫学

Outline of Final Research Achievements

In the case-cohort study, costly and laborious research can be greatly saved by measuring expensive genome information only from selected subjects, not the entire cohort. However, since data observed from the target frequently includes missing values, data analysis becomes difficult. In this study, we developed a novel theoretical framework for case-cohort studies when measurements of interests are known to be missing. We also examined the application of the proposed method to large-scale genomic epidemiological studies. From results verified by numerical experiments, when the outcomes of the majority of the subjects in the cohort are missing, the sub-cohort extraction or data analysis performed by the traditional method leads bias in the effects of the interesting variable.

Free Research Field

総合領域