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

A Study on Statistical Interpretation Methods for Machine Learning Results Using Shapley Values

Research Project

Project/Area Number 20K11938
Research Category

Grant-in-Aid for Scientific Research (C)

Allocation TypeMulti-year Fund
Section一般
Review Section Basic Section 61030:Intelligent informatics-related
Research InstitutionKumamoto University

Principal Investigator

NOHARA Yasunobu  熊本大学, 大学院先端科学研究部(工), 准教授 (30624829)

Co-Investigator(Kenkyū-buntansha) 松本 晃太郎  久留米大学, 付置研究所, 講師 (60932217)
Project Period (FY) 2020-04-01 – 2024-03-31
Project Status Completed (Fiscal Year 2023)
Budget Amount *help
¥4,160,000 (Direct Cost: ¥3,200,000、Indirect Cost: ¥960,000)
Fiscal Year 2023: ¥910,000 (Direct Cost: ¥700,000、Indirect Cost: ¥210,000)
Fiscal Year 2022: ¥910,000 (Direct Cost: ¥700,000、Indirect Cost: ¥210,000)
Fiscal Year 2021: ¥910,000 (Direct Cost: ¥700,000、Indirect Cost: ¥210,000)
Fiscal Year 2020: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Keywords機械学習 / 解釈手法 / シャプレー値 / 説明性の定量化 / 変数重要度 / 統計的仮説検定法 / 階層モデル
Outline of Research at the Start

近年、深層学習をはじめとする機械学習技術が注目され導入が進められつつあるが、なぜそのような結果が得られたかの説明・解釈性が強く求められている。本研究では、予測器の入力と出力の関係に注目してブラックボックス的な機械学習結果を解釈する手法を開発する。
開発手法では、経済学の分野で用いられている「複数人が協同した場合の利益の公平分配方法であるシャプレー値」を応用することで、数千にもおよぶ各説明変数が出力に及ぼす影響を線形和の形で適切に分離し、線形モデル等を前提として構築されている仮説検定や信頼区間といった統計学的な解釈手法をブラックボックスモデルに適用できるようにする。

Outline of Final Research Achievements

In recent years, machine learning technologies, including deep learning, have been gaining attention and are increasingly being implemented. However, there is a strong demand for the explanation and interpretability of the results these technologies produce. This study applies the Shapley value; a method of fair profit distribution among multiple collaborators used in economics; to the research of interpretability methods for machine learning models. First, we propose a method to quantitatively evaluate interpretability based on how accurately the model can be interpreted. Then, we theoretically prove that, in the absence of correlation among features, selecting features in descending order of the variance of their Shapley values maximizes interpretability when the number of usable features is limited.

Academic Significance and Societal Importance of the Research Achievements

既存の機械学習の解釈手法において、どの説明変数が重要であるかを表す指標である変数重要度は、経験的に使われてきたものであり、理論的な裏付けはなかった。本研究では、モデルをどれだけ正確に解釈できたかという説明性を定量的に評価する手法を提案し、その説明性を最大化するという理論的な裏付けがある手法を提案した点に大きな学術的意義を有する。
近年、機械学習は様々な分野で用いられようとしている。特に、病気の診断や自動運転等、間違いが重大な結果をもたらす分野において、機械学習がなぜそのような結果を出力したかを説明することは重要である。機械学習を広く社会へ適用するにあたって、本研究の社会的意義は大きい。

Report

(5 results)
  • 2023 Annual Research Report   Final Research Report ( PDF )
  • 2022 Research-status Report
  • 2021 Research-status Report
  • 2020 Research-status Report
  • Research Products

    (21 results)

All 2024 2023 2022 2021 2020

All Journal Article (13 results) (of which Peer Reviewed: 13 results,  Open Access: 8 results) Presentation (8 results) (of which Int'l Joint Research: 6 results)

  • [Journal Article] Development of Machine Learning Prediction Models for Self-Extubation After Delirium Using Emergency Department Data2024

    • Author(s)
      Matsumoto Koutarou、Nohara Yasunobu、Sakaguchi Mikako、Takayama Yohei、Yamashita Takanori、Soejima Hidehisa、Nakashima Naoki
    • Journal Title

      Studies in Health Technology and Informatics

      Volume: 2023 Pages: 1001-1005

    • DOI

      10.3233/shti231115

    • ISBN
      9781643684567, 9781643684574
    • Related Report
      2023 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Poststroke Clinical Outcomes in Comparison With Conventional Prognostic Scores: Multicenter, Hospital-Based Observational Study2024

    • Author(s)
      Fumi Irie, Koutarou Matsumoto, Ryu Matsuo, Yasunobu Nohara, Yoshinobu Wakisaka, Tetsuro Ago, Naoki Nakashima, Takanari Kitazono, Masahiro Kamouchi
    • Journal Title

      JMIR AI

      Volume: 3 Pages: e46840-e46840

    • DOI

      10.2196/46840

    • Related Report
      2023 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Temporal Generalizability of Machine Learning Models for Predicting Postoperative Delirium Using Electronic Health Record Data: Model Development and Validation Study2023

    • Author(s)
      Koutarou Matsumoto, Yasunobu Nohara, Mikako Sakaguchi, Yohei Takayama, Syota Fukushige, Hidehisa Soejima, Naoki Nakashima, Masahiro Kamouchi
    • Journal Title

      JMIR Perioperative Medicine

      Volume: 6 Pages: e50895-e50895

    • DOI

      10.2196/50895

    • Related Report
      2023 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Non-linear association between body weight and functional outcome after acute ischemic stroke2023

    • Author(s)
      Wakisaka Kayo、Matsuo Ryu、Matsumoto Koutarou、Nohara Yasunobu、Irie Fumi、Wakisaka Yoshinobu、Ago Tetsuro、Nakashima Naoki、Kamouchi Masahiro、Kitazono Takanari
    • Journal Title

      Scientific Reports

      Volume: 13 Issue: 1 Pages: 8697-8697

    • DOI

      10.1038/s41598-023-35894-y

    • Related Report
      2023 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Delirium Prediction Using Machine Learning Interpretation Method and Its Incorporation into a Clinical Workflow2023

    • Author(s)
      Matsumoto Koutarou、Nohara Yasunobu、Sakaguchi Mikako、Takayama Yohei、Fukushige Shota、Soejima Hidehisa、Nakashima Naoki
    • Journal Title

      Applied Sciences

      Volume: 13 Issue: 3 Pages: 1564-1564

    • DOI

      10.3390/app13031564

    • Related Report
      2022 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] Communications through contemporary tools of information and communication technology: Health among separated family members2022

    • Author(s)
      Nishikitani M, Ariyoshi M, Nohara Y, Umihara J
    • Journal Title

      JMIR Formative Research

      Volume: - Issue: 8 Pages: e34949-e34949

    • DOI

      10.2196/34949

    • Related Report
      2022 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] Explanation of Machine Learning Models Using Shapley Additive Explanation and Application for Real Data in Hospital2022

    • Author(s)
      Yasunobu Nohara, Koutarou Matsumoto, Hidehisa Soejima, Naoki Nakashima
    • Journal Title

      Computer Methods and Programs in Biomedicine

      Volume: 214 Pages: 106584-106584

    • DOI

      10.1016/j.cmpb.2021.106584

    • Related Report
      2021 Research-status Report
    • Peer Reviewed
  • [Journal Article] Machine Learning for Classification of Postoperative Patient Status Using Standardized Medical Data2022

    • Author(s)
      Takanori Yamashita, Yoshifumi Wakata, Hideki Nakaguma, Yasunobu Nohara, Shinj Hato, Susumu Kawamura, Shuko Muraoka, Masatoshi Sugita, Mihoko Okada, Naoki Nakashima, Hidehisa Soejima
    • Journal Title

      Computer Methods and Programs in Biomedicine

      Volume: 214 Pages: 106583-106583

    • DOI

      10.1016/j.cmpb.2021.106583

    • Related Report
      2021 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] Portable Health Clinic for Sustainable Care of Mothers and Newborns in Rural Bangladesh2021

    • Author(s)
      Kimiyo Kikuchi, Yoko Sato, Rieko Izukura, Mariko Nishikitani, Kiyoko Kato, Seiichi Morokuma, Meherun Nessa, Yasunobu Nohara, Fumihiko Yokota, Ashir Ahmed, Rafiqul Islam Maruf, Naoki Nakashima
    • Journal Title

      Computer Methods and Programs in Biomedicine

      Volume: 207 Pages: 106156-106156

    • DOI

      10.1016/j.cmpb.2021.106156

    • Related Report
      2021 Research-status Report
    • Peer Reviewed
  • [Journal Article] Association of serum bilirubin levels with risk of cancer development and total death2021

    • Author(s)
      Toyoshi Inoguchi, Yasunobu Nohara, Chinatsu Nojiri and Naoki Nakashima
    • Journal Title

      Scientific Reports

      Volume: 11 Issue: 1

    • DOI

      10.1038/s41598-021-92442-2

    • Related Report
      2021 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] 患者状態把握を目的とした機械学習と共起有向グラフによる診療プロセス解析2021

    • Author(s)
      山下 貴範, 若田 好史, 中熊 英貴, 野原 康伸, 岡田 美保子, 中島 直樹, 副島 秀久
    • Journal Title

      医療情報学

      Volume: 41 Pages: 29-37

    • Related Report
      2021 Research-status Report 2020 Research-status Report
    • Peer Reviewed
  • [Journal Article] A functional learning health system in Japan: Experience with processes and information infrastructure toward continuous health improvement2020

    • Author(s)
      Soejima Hidehisa、Matsumoto Koutarou、Nakashima Naoki、Nohara Yasunobu、Yamashita Takanori、Machida Jiro、Nakaguma Hideki
    • Journal Title

      Learning Health Systems

      Volume: e10252 Issue: 4 Pages: 1-12

    • DOI

      10.1002/lrh2.10252

    • Related Report
      2020 Research-status Report
    • Peer Reviewed
  • [Journal Article] Stroke Prognostic Scores and Data-Driven Prediction of Clinical Outcomes After Acute Ischemic Stroke2020

    • Author(s)
      Matsumoto Koutarou、Nohara Yasunobu、Soejima Hidehisa、Yonehara Toshiro、Nakashima Naoki、Kamouchi Masahiro
    • Journal Title

      Stroke

      Volume: 51 Issue: 5 Pages: 1477-1483

    • DOI

      10.1161/strokeaha.119.027300

    • Related Report
      2020 Research-status Report
    • Peer Reviewed
  • [Presentation] Development of Machine Learning Prediction Models for Self-Extubation After Delirium Using Emergency Department Data2023

    • Author(s)
      Matsumoto Koutarou、Nohara Yasunobu、Sakaguchi Mikako、Takayama Yohei、Yamashita Takanori、Soejima Hidehisa、Nakashima Naoki
    • Organizer
      MedInfo2023
    • Related Report
      2023 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Developing a Learning Health System for Delirium using XAI2022

    • Author(s)
      Koutarou Matsumoto, Yasunobu Nohara, Mikako Sakaguchi, Yohei Takayama, Hidehisa Soejima and Naoki Nakashima
    • Organizer
      12th Biennial Conference of the Asia Pacific Association for Medical Informatics
    • Related Report
      2022 Research-status Report
    • Int'l Joint Research
  • [Presentation] Explanation of Machine Learning Models Using SHAP Considering Interaction Effects2022

    • Author(s)
      Yasunobu Nohara, Toyoshi Inoguchi, Chinatsu Nojiri and Naoki Nakashima
    • Organizer
      2nd ICML Workshop on Interpretable Machine Learning in Healthcare
    • Related Report
      2022 Research-status Report
    • Int'l Joint Research
  • [Presentation] 交互作用を考慮したSHAPによる機械学習モデルの解釈2022

    • Author(s)
      野原 康伸, 井口 登與志, 野尻 千夏, 中島 直樹
    • Organizer
      第26回日本医療情報学会春季学術大会
    • Related Report
      2022 Research-status Report
  • [Presentation] Predictors of Intracerebral Hematoma Enlargement Using Brain CT Images in Emergency Medical Care2021

    • Author(s)
      Kazunori Oka, Takumi Hirahara, Yasunobu Nohara, Sozo Inoue, Koichi Arimura, Syoji Kobashi and Koji Iihara
    • Organizer
      5th IEEE International Conference on Cybernetics
    • Related Report
      2021 Research-status Report
    • Int'l Joint Research
  • [Presentation] 交互作用を考慮したSHAPによる機械学習モデルの解釈手法の提案2021

    • Author(s)
      野原 康伸, 井口 登興志, 野尻 千夏, 中島 直樹
    • Organizer
      第41回医療情報学連合大会
    • Related Report
      2021 Research-status Report
  • [Presentation] Explanation of Machine Learning Models Using Shapley Additive Explanation and Application for Real Data in Hospital2020

    • Author(s)
      Yasunobu Nohara, Koutarou Matsumoto, Hidehisa Soejima, Naoki Nakashima
    • Organizer
      11th Biennial Conference of the Asia Pacific Association for Medical Informatics
    • Related Report
      2020 Research-status Report
    • Int'l Joint Research
  • [Presentation] Machine Learning for Classification of Postoperative Patient Status Using Standardized Medical Data2020

    • Author(s)
      Takanori Yamashita, Yoshifumi Wakata, Hideki Nakaguma, Yasunobu Nohara, Shinji Hato, Susumu Kawamura, Shuko Muraoka, Masatoshi Sugita, Mihoko Okada, Naoki Nakashima, Hidehisa Soejima
    • Organizer
      11th Biennial Conference of the Asia Pacific Association for Medical Informatics
    • Related Report
      2020 Research-status Report
    • Int'l Joint Research

URL: 

Published: 2020-04-28   Modified: 2025-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi