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

2023 Fiscal Year Final Research Report

Identification of prognostic factors for elderly hospitalized patients with multiple diseases: improving interpretability of machine learning models.

Research Project

  • PDF
Project/Area Number 18K18471
Research Category

Grant-in-Aid for Challenging Research (Exploratory)

Allocation TypeMulti-year Fund
Review Section Studies on the Super-Aging Society
Research InstitutionYokohama City University (2020, 2022-2023)
Institute for Health Economics and Policy, Association for Health Economics Rsearch and Social Insurance and Welfare (2018-2019)

Principal Investigator

Shimizu Sayuri  横浜市立大学, データサイエンス研究科, 講師 (60625408)

Co-Investigator(Kenkyū-buntansha) 原 聡  大阪大学, 産業科学研究所, 准教授 (40780721)
伏見 清秀  東京医科歯科大学, 大学院医歯学総合研究科, 教授 (50270913)
Project Period (FY) 2018-06-29 – 2024-03-31
Keywords医療データベース / 機械学習
Outline of Final Research Achievements

The increasing number of elderly individuals with multiple diseases and reduced physical resilience makes it imperative to evaluate them from a comprehensive perspective in database studies for clinical assessment. In this study, we constructed prediction models using multiple methods, including conventional models, gradient boosting models, and models that take interpretability into account, suggesting that it is possible to improve the accuracy of prediction models. This analysis reiterates the significance of selecting an analytical model that accounts for the distinctive characteristics of healthcare administrative data, the analytical compatibility with machine learning models, and interpretability.

Free Research Field

ヘルスサービスリサーチ

Academic Significance and Societal Importance of the Research Achievements

臨床現場から日々生成される医療データが蓄積され、世界的な潮流として、これらのデータを臨床や政策に活用しようという動きが広がっています。加えて、従来型の統計モデルから機械学習モデルへのシフトがおこっており、これらのモデルを医療管理分野の分析にどのように活かすかが課題となっていました。本研究では、機械学習モデルがより精度高く予測可能でありましたが、解釈可能性に留意する必要があることが示唆されました。

URL: 

Published: 2025-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi