2018 Fiscal Year Final Research Report

Novel strategy for intensive care by using application of artificial intelligence to data mining of Diagnosis Procedure Combination payment system

Research Project

PDF

Project/Area Number	17K17607
Research Category	Grant-in-Aid for Young Scientists (B)
Allocation Type	Multi-year Fund
Research Field	Medical and hospital managemen Emergency medicine
Research Institution	Tohoku University
Principal Investigator	Shiga Takuya 東北大学, 大学病院, 助教 (90539074)
Research Collaborator	IKUMI saori KOBAYASHI naoya SUGINO shigekazu SAITO koji NAKAYAMA masaharu FUJIMORI kenji YAMAUCHI masanori
Project Period (FY)	2017-04-01 – 2019-03-31
Keywords	集中治療 / コスト / 重症度 / ビッグデータ / DPC
Outline of Final Research Achievements	I constructed integrated big data in the intensive care unit (ICU) of Tohoku University Hospital, analyzed the relationship between the severity and the cost of each patient in the ICU. I planned to develop the original algorithm to derive the optimal treatments that would achieve optimal medical resource distribution and maximum medical benefits, depending on the changes in each patient's condition of severity. I set up a server to construct panel database. In regards to the patient’s severity, I gathered the ICU system data, and bed-side monitoring data of each patients admitted to our ICU for the past five years. In regards to the cost, I gathered material cost, labor cost, expense data, medical record and DPC/PDPS data with the cooperation of the university hospital medical affairs department. I cleaned and integrated the all above data, transferred to my server, and constructed the panel database. Finally, I calculated departmental costing and cost forecasting with this database.
Free Research Field	医療管理学
Academic Significance and Societal Importance of the Research Achievements	本研究統合ビッグデータを用いて当院集中治療室における原価計算を行った。この計算によると、2017年単月において、費用を除いた材料費と労務費の合計が75,710,585円とすでに収入74,748,228を上回っており、費用を加味すると大幅な赤字であることが判明した。この結果の一部は第46回日本集中治療医学会学術集会にて報告した。また、重症度とコスト予測の関連において、看護必要度に重症度スコアであるSOFAスコアを加えると、コスト予測精度が上がる傾向があることが分かった。この統合データベースは今後も運用可能であるため、さらに深層学習による1日ごとの重症度、医療コスト、診療報酬額の解析を継続する。