Heuristic research of ICU big data : Cost beneficial intensive care treatment with artificial intelligence

2022 Fiscal Year Final Research Report

• Project/Area Number: 19K18311
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 55060:Emergency medicine-related
• Research Institution: Tohoku University
• Principal Investigator: Shiga Takuya 東北大学, 大学病院, 講師 (90539074)
• Project Period (FY): 2019-04-01 – 2023-03-31
• Keywords: 集中治療 / コスト / データベース / 原価計算 / ビッグデータ

Outline of Final Research Achievements

In this study, we first collected data on patients admitted to the ICU of the hospital. We performed departmental costing of the intensive care unit in the hospital. The data was machine-learned, and the physiological severity of illness was a better predictor of costs than the nursing necessity for calculation of the specific intensive care unit management fee.
Next, machine learning of data on patients admitted to ICUs nationwide from the MHLW Fushimi Group DPC database suggested that the SOFA score was effective in predicting costs. In addition, a comparison of the ICU Management Fee 1 and 2, which requires trained professional staff, and the ICU Management Fee 3 and 4, which does not, showed that the Management Fee 1 and 2 was associated with a lower patient mortality rate and was more cost-effective than the Management Fee 3 and 4.

Free Research Field

集中治療

Academic Significance and Societal Importance of the Research Achievements

本研究は，本邦における集中治療室の原価計算情報がほとんどないため，その基礎情報となる．生理学的重症度スコアが，本邦の医療制度において医療コストが関連することを示した．また，厚生労働科学研究伏見班DPCデータベースを用いて，層別化されている特定集中治療室管理料1，2と3，4の比較と費用対効果分析を行った．より手厚いスタッフィングを行っている特定集中治療室管理料1，2は，3，4よりも死亡率が低く，費用対効果にも優れていることを示した．したがって，本研究成果は，今後の集中治療における医療費のバランスをいかに取るか，患者層別化の方法と費用対効果についての情報として，大きな社会的意義がある．