2016 Fiscal Year Final Research Report
Chort based Indivualized Risk Estimation By biomarkers and Machine Learning
Project/Area Number |
26460790
|
Research Category |
Grant-in-Aid for Scientific Research (C)
|
Allocation Type | Multi-year Fund |
Section | 一般 |
Research Field |
Epidemiology and preventive medicine
|
Research Institution | National Cardiovascular Center Research Institute |
Principal Investigator |
Kunihiro Nishimura 国立研究開発法人国立循環器病研究センター, 研究開発基盤センター, 室長 (70397834)
|
Project Period (FY) |
2014-04-01 – 2017-03-31
|
Keywords | コホート研究 / リスク予測 / 機械学習 / 虚血性心疾患 |
Outline of Final Research Achievements |
We deveoped coronary heart disease (CHD) risk score, called Suita Score based on 10 years follo-up results from Suita study for praimary prevention. We also devoped the risk score for major cardiac disease( MACE) prediction score based on the follow-up data of 1920 prospective observaional cases with cariovascular diseases using machine learnig. We adopted ensemble learning with support vectors and random forests to predict 1 year MACE in the socondary prevention using around 40 rountinely used biomarkers in daily practices. The total acuracy was around 90% and it was reproudcible in validation data sets, with more precise prediction ability than existing MACE prediction model, such as GRACE socre.
|
Free Research Field |
循環器疫学、医学統計学
|