2022 Fiscal Year Final Research Report
Innovative Ventilator Weaning Strategy Utilizing Artificial Intelligence and Intensive Care Patient Information Systems
Project/Area Number |
20K17876
|
Research Category |
Grant-in-Aid for Early-Career Scientists
|
Allocation Type | Multi-year Fund |
Review Section |
Basic Section 55060:Emergency medicine-related
|
Research Institution | Nippon Medical School |
Principal Investigator |
|
Project Period (FY) |
2020-04-01 – 2023-03-31
|
Keywords | 人工知能 / 人工呼吸器 / 集中治療 / 抜管 / 機械学習 |
Outline of Final Research Achievements |
Our study aimed at reducing extubation failures and minimizing the duration of mechanical ventilation through AI, and was conducted in the following three phases: (1) Exploration: Using data extracted from the Intensive Care Patient Information System, we assessed the feasibility of predicting the success or failure of extubation. If predictable, we evaluated the significance of various features and their medical validity. (2) Accuracy Improvement: We investigated whether the accuracy of AI predictions for extubation success or failure could be enhanced in comparison with previous studies. (3) Implementation: We examined the possibility of real-time predictions via AI, with an eye toward clinical research implementation. As a result of these efforts, we have published one original research paper and one review paper in English.
|
Free Research Field |
救急医学
|
Academic Significance and Societal Importance of the Research Achievements |
患者背景、バイタルサイン、検査データ、人工呼吸器のデータなど、57の特徴に関する情報を抽出し、人工呼吸管理が必要であるか否かのラベルを付け、3つの学習アルゴリズムを用いて、抜管の予測モデルを作成した。また、精度を向上させるべく、不確実性を考慮したニューラルネットワークモデルを作成した。最後に入院中の患者データを利用して予測ができるよう実装を行った。
|