Knowledge and data-driven prediction of organ failure in critical care patients.

IF 4.7 3区医学 Q1 MEDICAL INFORMATICS Health Information Science and Systems Pub Date : 2023-01-23 eCollection Date: 2023-12-01 DOI:10.1007/s13755-023-00210-5

Xinyu Ma, Meng Wang, Sihan Lin, Yuhao Zhang, Yanjian Zhang, Wen Ouyang, Xing Liu

{"title":"Knowledge and data-driven prediction of organ failure in critical care patients.","authors":"Xinyu Ma, Meng Wang, Sihan Lin, Yuhao Zhang, Yanjian Zhang, Wen Ouyang, Xing Liu","doi":"10.1007/s13755-023-00210-5","DOIUrl":null,"url":null,"abstract":"Purpose: The early detection of organ failure mitigates the risk of post-intensive care syndrome and long-term functional impairment. The aim of this study is to predict organ failure in real-time for critical care patients based on a data-driven and knowledge-driven machine learning method (DKM) and provide explanations for the prediction by incorporating a medical knowledge graph.Methods: The cohort of this study was a subset of the 4,386 adult Intensive Care Unit (ICU) patients from the MIMIC-III dataset collected between 2001 and 2012, and the primary outcome was the Delta Sequential Organ Failure Assessment (SOFA) score. A real-time Delta SOFA score prediction model was developed with two key components: an improved deep learning temporal convolutional network (S-TCN) and a graph-embedding feature extraction method based on a medical knowledge graph. Entities and relations related to organ failure were extracted from the Unified Medical Language System to build the medical knowledge graph, and patient data were mapped onto the graph to extract the embeddings. We measured the performance of our DKM approach with cross-validation to avoid the formation of biased assessments.Results: An area under the receiver operating characteristic curve (AUC) of 0.973, a precision of 0.923, a NPV of 0.989, and an F1 score of 0.927 were achieved using the DKM approach, which significantly outperformed the baseline methods. Additionally, the performance remained stable following external validation on the eICU dataset, which consists of 2,816 admissions (AUC = 0.981, precision = 0.860, NPV = 0.984). Visualization of feature importance for the Delta SOFA score and their relationships on the basic clinical medical (BCM) knowledge graph provided a model explanation.Conclusion: The use of an improved TCN model and a medical knowledge graph led to substantial improvement in prediction accuracy, providing generalizability and an independent explanation for organ failure prediction in critical care patients. These findings show the potential of incorporating prior domain knowledge into machine learning models to inform care and service planning.Supplementary information: The online version of this article contains supplementary material available 10.1007/s13755-023-00210-5.","PeriodicalId":46312,"journal":{"name":"Health Information Science and Systems","volume":null,"pages":null},"PeriodicalIF":4.7000,"publicationDate":"2023-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9871106/pdf/","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Health Information Science and Systems","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1007/s13755-023-00210-5","RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2023/12/1 0:00:00","PubModel":"eCollection","JCR":"Q1","JCRName":"MEDICAL INFORMATICS","Score":null,"Total":0}

引用次数: 1

Abstract

Purpose: The early detection of organ failure mitigates the risk of post-intensive care syndrome and long-term functional impairment. The aim of this study is to predict organ failure in real-time for critical care patients based on a data-driven and knowledge-driven machine learning method (DKM) and provide explanations for the prediction by incorporating a medical knowledge graph.

Methods: The cohort of this study was a subset of the 4,386 adult Intensive Care Unit (ICU) patients from the MIMIC-III dataset collected between 2001 and 2012, and the primary outcome was the Delta Sequential Organ Failure Assessment (SOFA) score. A real-time Delta SOFA score prediction model was developed with two key components: an improved deep learning temporal convolutional network (S-TCN) and a graph-embedding feature extraction method based on a medical knowledge graph. Entities and relations related to organ failure were extracted from the Unified Medical Language System to build the medical knowledge graph, and patient data were mapped onto the graph to extract the embeddings. We measured the performance of our DKM approach with cross-validation to avoid the formation of biased assessments.

Results: An area under the receiver operating characteristic curve (AUC) of 0.973, a precision of 0.923, a NPV of 0.989, and an F1 score of 0.927 were achieved using the DKM approach, which significantly outperformed the baseline methods. Additionally, the performance remained stable following external validation on the eICU dataset, which consists of 2,816 admissions (AUC = 0.981, precision = 0.860, NPV = 0.984). Visualization of feature importance for the Delta SOFA score and their relationships on the basic clinical medical (BCM) knowledge graph provided a model explanation.

Conclusion: The use of an improved TCN model and a medical knowledge graph led to substantial improvement in prediction accuracy, providing generalizability and an independent explanation for organ failure prediction in critical care patients. These findings show the potential of incorporating prior domain knowledge into machine learning models to inform care and service planning.

Supplementary information: The online version of this article contains supplementary material available 10.1007/s13755-023-00210-5.

查看原文

微信好友朋友圈 QQ好友复制链接

本刊更多论文

重症监护患者器官衰竭的知识和数据驱动预测。

目的：早期发现器官衰竭可降低重症监护后综合征和长期功能损害的风险。本研究的目的是基于数据驱动和知识驱动的机器学习方法（DKM）实时预测重症监护患者的器官衰竭，并通过结合医学知识图为预测提供解释。方法：本研究的队列是2001年至2012年间收集的MIMIC-III数据集中4386名成人重症监护室（ICU）患者的子集，主要结果是德尔塔顺序器官衰竭评估（SOFA）评分。开发了一个实时Delta SOFA分数预测模型，该模型由两个关键组件组成：改进的深度学习时间卷积网络（S-TCN）和基于医学知识图的图嵌入特征提取方法。从统一医学语言系统中提取与器官衰竭相关的实体和关系以构建医学知识图，并将患者数据映射到图上以提取嵌入。我们通过交叉验证来衡量DKM方法的性能，以避免形成有偏见的评估。结果：使用DKM方法获得了0.973的受试者工作特征曲线下面积（AUC）、0.923的精度、0.989的NPV和0.927的F1分数，显著优于基线方法。此外，在eICU数据集上进行外部验证后，性能保持稳定，该数据集包括2816例入院（AUC = 0.981，精度 = 0.860，NPV = 0.984）。德尔塔SOFA评分的特征重要性及其在基础临床医学（BCM）知识图上的关系的可视化提供了模型解释。结论：使用改进的TCN模型和医学知识图显著提高了预测准确性，为重症监护患者的器官衰竭预测提供了可推广性和独立解释。这些发现显示了将先验领域知识纳入机器学习模型以为护理和服务规划提供信息的潜力。补充信息：本文的在线版本包含补充材料10.1007/s13755-023-00210-5。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文去求助

来源期刊

Health Information Science and Systems MEDICAL INFORMATICS-

CiteScore

11.30

自引率

5.00%

发文量

期刊介绍： Health Information Science and Systems is a multidisciplinary journal that integrates artificial intelligence/computer science/information technology with health science and services, embracing information science research coupled with topics related to the modeling, design, development, integration and management of health information systems, smart health, artificial intelligence in medicine, and computer aided diagnosis, medical expert systems. The scope includes: i.) smart health, artificial Intelligence in medicine, computer aided diagnosis, medical image processing, medical expert systems ii.) medical big data, medical/health/biomedicine information resources such as patient medical records, devices and equipments, software and tools to capture, store, retrieve, process, analyze, optimize the use of information in the health domain, iii.) data management, data mining, and knowledge discovery, all of which play a key role in decision making, management of public health, examination of standards, privacy and security issues, iv.) development of new architectures and applications for health information systems.