主管:国家卫生健康委员会
主办:国家卫生健康委医院管理研究所
中国科技核心期刊(中国科技论文统计源期刊)
中国科学引文数据库(CSCD)核心库期刊
《中文核心期刊要目总览》核心期刊

中国护理管理 ›› 2026, Vol. 26 ›› Issue (10): 1540-1545.doi: 10.3969/j.issn.1672-1756.2026.10.019

• 数智化护理 • 上一篇    下一篇

危重症患者谵妄管理临床决策支持系统的开发与可用性评价

张山 刘璐 张艺丹 丁舒 王军 吴瑛   

  1. 首都医科大学护理学院,100069 北京市(张山,刘璐,张艺丹,丁舒,吴瑛);首都医科大学附属北京朝阳医院护理部(丁舒);首都医科大学宣武医院神经外科(王军)
  • 出版日期:2026-10-15 发布日期:2026-10-15
  • 通讯作者: 吴瑛,博士,教授,临床护理学院院长,E-mail:helenywu@vip.163.com
  • 作者简介:张山,博士,副教授,教学秘书
  • 基金资助:
    国家自然科学基金青年项目(72304196)

Development and usability evaluation of a clinical decision support system for delirium management in critically ill patients

ZHANG Shan, LIU Lu, ZHANG Yidan, DING Shu, WANG Jun, WU Ying   

  1. School of Nursing, Capital Medical University, Beijing, 100069, China
  • Online:2026-10-15 Published:2026-10-15
  • Contact: E-mail:helenywu@vip.163.com

摘要: 目的:构建基于自适应机制的危重症患者谵妄管理临床决策支持系统,并评价其可用性,为医护人员获取最新决策支持提供辅助工具。方法:构建危重症患者谵妄指南知识图谱,并基于自适应机制开发危重症患者谵妄管理临床决策支持系统,采用图注意力网络将患者数据与知识图谱进行融合。采用便利抽样法,于2024年3月至2025年4月纳入来自北京市2家三级甲等医院的962例ICU患者。按7∶2∶1分成训练集(673例)、测试集(192例)和验证集(97例),分别用于模型参数训练、推荐性能评价、超参数调整与早停。采用卡方检验比较静态知识图谱模式和自适应机制模式下的护理措施推荐准确性。2025年7月—12月,便利选取上述两家医院的87名ICU护士,采用“智能护理系统可用性评价问卷”进行可用性测试。结果:自适应机制模式下系统推荐准确率为86.5%,高于静态知识图谱模式的73.4%(P<0.05)。可用性评价总分为(187.52±24.55)分,其中感知有用性为(40.82±5.37)分、感知易用性为(75.25±9.95)分、使用态度为(22.26±3.10)分、使用倾向为(31.44±4.32)分、系统远期影响为(17.76± 2.65)分。结论:基于自适应机制的危重症患者谵妄管理临床决策支持系统能够准确推荐护理措施,并具有良好的系统可用性,可为护理决策提供支持。

关键词: 谵妄管理;知识图谱;临床决策支持系统;自适应机制

Abstract: Objective: To construct a clinical decision support system for delirium management in critically ill patients based on an adaptive mechanism, and to evaluate its clinical usability, providing an auxiliary tool for healthcare professionals to obtain the latest decision support. Methods: A knowledge graph for delirium guidelines in critically ill patients was constructed, and a clinical decision support system for delirium management in critically ill patients was developed based on an adaptive mechanism. The graph attention network was used to integrate patient data with the knowledge graph. Data of 962 ICU patients from two tertiary grade A hospitals in Beijing selected through convenience sampling were included from March 2024 to April 2025 and were divided into a training set (673 cases), a test set (192 cases), and a validation set (97 cases) at a ratio of 7?:?2?:?1, which were used for model parameter training, recommendation performance evaluation, hyperparameter adjustment and early stopping, respectively. The chi-square test was used to compare the accuracy of nursing intervention recommendations under the static knowledge graph mode and the adaptive mechanism mode. From July to December 2025, the Intelligent Care System Usability Evaluation Questionnaire was used to conduct usability tests among 87 ICU nurses recruited via convenience sampling from the above two tertiary hospitals. Results: The accuracy rate of system recommendations in the adaptive mechanism mode was 86.5%, which was higher than 73.4% in the static knowledge graph mode (P<0.05). The total score of usability evaluation was 187.52±24.55, including perceived usefulness 40.82±5.37, perceived ease of use 75.25±9.95, attitude toward use 22.26±3.10, behavioral intention to use 31.44±4.32, and long-term impact of the system 17.76±2.65. Conclusion: The clinical decision support system based on the adaptive mechanism for delirium management in critically ill patients can accurately recommend nursing interventions and exhibits favorable usability, which can provide support for clinical nursing decision-making.

Key words: delirium management; knowledge graph; clinical decision-making support system; adaptive mechanism

中图分类号:  R47;R197