以物联网数据为核心的围手术期临床专科数据资源模型研究与实践

Research and Practice on Perioperative Clinical Specialty Data Resource Model Centered on Internet of Things Data

  • 摘要: 摘要:背景 围手术期真实世界数据(RWD)虽涉及详细的患者信息和临床诊疗过程,却很少能被用于临床研究。目的 本研究通过构建围手术期数据资源模型体系(WIRE),整合手术患者围手术期的多源RWD,推动智能化技术真实世界研究的开展与应用。方法 研究团队基于HL7 参考信息模型(RIM),结合医疗物联网数据特征和临床数据模型,设计了WIRE体系,实现了围手术期医学信息的数据模型层整合。同时,研究了基于WIRE的围手术期智能化预警技术的真实世界研究(RWS)通用方法,并针对术中低氧血症和低血压开展预警模型研究,研发集成预警模型的手术患者风险预警系统。结果 在解放军总医院第六医学中心和广东省人民医院均成功构建了基于WIRE的围手术期专科数据资源库,分别汇集了6 483 台次、27 939 台次手术相关数据,为预警模型提供了充足的数据资源。术中低血压预警模型以及术中低氧血症在提前3分钟和提前5 分钟预测模型的精确率、召回率、F1 分数均优于麻醉医师表现(P<0. 05)。此外,开发的风险预警系统实现了对手术患者术中并发症发生可能性的即时预警。结论 WIRE能有效整合手术患者围手术期RWD,促进临床科研工作开展,并对临床数据要素的产业化价值释放提供了实践参考。

     

    Abstract: Abstract: Background Perioperative real-world data (RWD), despite its comprehensive capture of patient information and clinical treatment processes, is seldom utilized in clinical research. Objective This study aims to construct a perioperative data resource model system (WIRE) to integrate multi-source RWD during the perioperative period and to advance the application of intelligent technology in real-world studies.Methods The research team, grounded in the HL7 Reference Information Model (RIM) and incorporating characteristics of medical Internet of Things data along with clinical data models, designed the WIRE system to achieve integration at the data model layer of perioperative medical information. Concurrently, the team explored general methods for real-world studies (RWS) of perioperative intelligent early warning technology based on WIRE and developed early warning models for intraoperative hypoxemia and hypotension, culminating in the creation of a risk early warning system integrated with these models.Results The perioperative specialty data resource libraries based on WIRE were successfully established at the Sixth Medical Center of the General Hospital of the People's Liberation Army and Guangdong Provincial People's Hospital, respectively aggregating data from 6, 483 and 27, 939 surgical cases, providing ample data resources for the early warning models. The intraoperative hypotension early warning model and the intraoperative hypoxemia prediction models for 3-minute and 5-minute early warnings have a higher accuracy, recall, and F1 scores compared to the performance of anesthesiologists(P<0.05). Moreover, the developed risk early warning system enabled real-time alerting for the potential occurrence of intraoperative complications in surgical patients.Conclusion WIRE can effectively integrate RWD from the perioperative period of surgical patients, facilitating the advancement of clinical research and providing practical references for the industrial application of clinical data elements.

     

/

返回文章
返回