• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Predictive model of cerebral infarction of atrial fibrillation by deep learning

Research Project

Project/Area Number 16K09419
Research Category

Grant-in-Aid for Scientific Research (C)

Allocation TypeMulti-year Fund
Section一般
Research Field Cardiovascular medicine
Research InstitutionInternational University of Health and Welfare (2017-2018)
The University of Tokyo (2016)

Principal Investigator

Sakurai Ryota  国際医療福祉大学, 医学部, 教授 (80466747)

Project Period (FY) 2016-04-01 – 2019-03-31
Project Status Completed (Fiscal Year 2018)
Budget Amount *help
¥2,990,000 (Direct Cost: ¥2,300,000、Indirect Cost: ¥690,000)
Fiscal Year 2018: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2017: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2016: ¥910,000 (Direct Cost: ¥700,000、Indirect Cost: ¥210,000)
Keywords機械学習 / 深層学習 / 予測モデル / 電子的診療データ / 心房細動 / 脳梗塞発症 / 臨床 / データベース
Outline of Final Research Achievements

The purpose of this study was to compare the models using deep learning and those with conventional machine learning for predicting future cerebral infarction in patients with atrial fibrillation using a large-scale electronic medical data.
With medical data of thousands of cases and tens of thousands of features in a single hospital stored in SS-MIX2 storage, an optimal prediction algorithm could be constructed using gradient boosting model, by which test accuracy, precision, and recall achieved 1.000.

Academic Significance and Societal Importance of the Research Achievements

大規模な電子的診療データが利用できるようになり、新しい機械学習の手法である深層学習を用いると、より正確に疾患の発症を予測できることが期待されるが、専用のコンピュータや多くのハードディスク容量、消費電力、計算時間を要するなど、欠点もある。今後さらなる改良が期待されるが、現時点で利用可能な計算機や機械学習の手法を用いることでも、実際に自分が治療を受けている医療機関でのデータのみで、十分正確な疾患の発症が予測できることが明らかとなった。

Report

(4 results)
  • 2018 Annual Research Report   Final Research Report ( PDF )
  • 2017 Research-status Report
  • 2016 Research-status Report

URL: 

Published: 2016-04-21   Modified: 2021-12-27  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi