基于机器学习的心源性休克患者院内死亡的预测研究
Research on Prediction Model of Cardiogenic Shock in-Hospital Death Based on Machine Learning
DOI: 10.12677/ACM.2022.12111454, PDF,   
作者: 周 弛, 廉哲勋*：青岛大学附属医院心血管内科，山东 青岛
关键词: 机器学习；预测模型；心源性休克；Machine Learning； Prediction Model； Cardiogenic Shock
摘要: 目的：构建心源性休克患者院内死亡预测模型。方法：利用MIMIC-III数据库，收集人口特征、实验室检查、合并症等87个指标并进行特征选择后，使用随机森林、Logistic回归、XGBoost、卷积神经网络算法，构建预测模型。用敏感性、特异性、曲线下面积、准确性来比较这4种模型的性能。结果：在这项研究的804名患者中，有209名患者(26%)出现院内死亡。在预测院内死亡时，四个模型的接收器工作特征曲线(ROC)的曲线下面积(AUC)在0.757至0.848的范围内。在所有模型中，XGBoost的灵敏度最高(87.3%)，特异性(81%)和准确性最高(84.6%)。结论：机器学习算法可以准确预测心源性休克患者院内死亡率，尤其是XGBoost模型。
Abstract: Objective: To construct a prediction model for in-hospital death in patients with cardiogenic shock. Methods: Using the MIMIC-III database, 87 indicators of demographic characteristics, laboratory tests, and comorbidities were collected and then feature selection. Logistic regression, random for-est, XGBoost, and convolutional neural network algorithms were employed to build models. Sensi-tivity, specificity, accuracy, and area under the curve were applied to access the performance. Re-sult: Among 804 patients enrolled in this study, 209 patients (26%) died in hospital. In the predic-tion of the in-hospital death, the areas under the curve (AUCs) of the receiver operating characteris-tic curves (ROCs) of the four models ranged from 0.757 to 0.848. Among all models, XGBoost achieved the highest sensitivity (87.3%), specificity (81%) and accuracy (84.6%). Conclusion: Ma-chine learning algorithms can accurately predict the in-hospital mortality of patients with cardio-genic shock, especially the XGBoost model.
文章引用：周弛, 廉哲勋. 基于机器学习的心源性休克患者院内死亡的预测研究[J]. 临床医学进展, 2022, 12(11): 10081-10090. https://doi.org/10.12677/ACM.2022.12111454

参考文献

[1] Levy, B., Bastien, O., Karim, B., et al. (2015) Experts’ Recommendations for the Management of Adult Patients with Cardiogenic Shock. Annals of Intensive Care, 5, Article No. 52. [Google Scholar] [CrossRef] [PubMed]
[2] Zannad, F., Mebazaa, A., Juillière, Y., et al. (2006) Clinical Pro-file, Contemporary Management and One-Year Mortality in Patients with Severe Acute Heart Failure Syndromes: The EFICA Study. European Journal of Heart Failure, 8, 697-705. [Google Scholar] [CrossRef] [PubMed]
[3] Pöss, J., Köster, J., Fuernau, G., et al. (2017) Risk Stratification for Patients in Cardiogenic Shock after Acute Myocardial Infarction. Journal of the American College of Cardiology, 69, 1913-1920. [Google Scholar] [CrossRef] [PubMed]
[4] 李昊朋. 基于机器学习方法的智能机器人探究[J]. 通讯世界, 2019, 26(4): 241-242.
[5] Johnson, A.E., Pollard, T.J., Shen, L., et al. (2016) MIMIC-III, a Freely Accessible Critical Care Database. Scientific Data, 3, Article ID: 160035. [Google Scholar] [CrossRef] [PubMed]
[6] Johnson, A.E., Stone, D.J., Celi, L.A. and Pollard, T.J. (2018) The MIMIC Code Repository: Enabling Reproducibility in Critical Care Research. Journal of the American Medical Informatics Association, 25, 32-39. [Google Scholar] [CrossRef] [PubMed]
[7] 周昀锴. 机器学习及其相关算法简介[J]. 科技传播, 2019, 11(6): 153-154, 165.
[8] Lundberg, S.M. and Lee, S.-I. (2017) A Unified Approach to Interpreting Model Predictions. In: Proceedings of the 31st International Conference on Neural Information Processing Systems, Curran Associates Inc., Long Beach, 4768-4777.
[9] Abdin, A., Pöss, J., Fuernau, G., et al. (2018) Prognostic Impact of Baseline Glucose Levels in Acute Myocardial Infarction Complicated by Cardiogenic Shock—A Sub-Study of the IABP-SHOCK II-Trial [Corrected]. Clinical Research in Cardiology, 107, 517-523. [Google Scholar] [CrossRef] [PubMed]
[10] Shin, S., Austin, P.C., Ross, H.J., et al. (2021) Machine Learning vs. Conventional Statistical Models for Predicting Heart Failure Readmission and Mortality. ESC Heart Fail, 8, 106-115. [Google Scholar] [CrossRef] [PubMed]
[11] D’Ascenzo, F., De Filippo, O., Gallone, G., et al. (2021) Machine Learn-ing-Based Prediction of Adverse Events Following an Acute Coronary Syndrome (PRAISE): A Modelling Study of Pooled Datasets. The Lancet, 397, 199-207. [Google Scholar] [CrossRef] 
[12] Chung, M., Zhao, Y., Strom, J.B., Shen, C. and Yeh, R.W. (2019) Extracorporeal Membrane Oxygenation Use in Cardiogenic Shock: Impact of Age on In-Hospital Mortality, Length of Stay, and Costs. Critical Care Medicine, 47, e214-e221. [Google Scholar] [CrossRef] 
[13] Geppert, A., Dorninger, A., Delle-Karth, G., Zorn, G., Heinz, G. and Huber, K. (2006) Plasma Concentrations of Interleukin-6, Organ Failure, Vasopressor Support, and Suc-cessful Coronary Revascularization in Predicting 30-Day Mortality of Patients with Cardiogenic Shock Complicating Acute Myocardial Infarction. Critical Care Medicine, 34, 2035-2042. [Google Scholar] [CrossRef] 
[14] Théroux, P., Armstrong, P.W., Mahaffey, K.W., et al. (2005) Prognostic Significance of Blood Markers of Inflammation in Patients with ST-Segment Elevation Myocardial Infarction Undergoing Primary Angioplasty and Effects of Pexelizumab, a C5 Inhibitor: A Sub-Study of the COMMA Trial. European Heart Journal, 26, 1964-1970. [Google Scholar] [CrossRef] [PubMed]
[15] Kim, C.H., Park, J.T., Kim, E.J., et al. (2013) An Increase in Red Blood Cell Distribution Width from Baseline Predicts Mortality in Patients with Severe Sepsis or Septic Shock. Critical Care, 17, R282. [Google Scholar] [CrossRef] [PubMed]
[16] Wang, B., Aihemaiti, G., Cheng, B. and Li, X. (2019) Red Blood Cell Distri-bution Width Is Associated with All-Cause Mortality in Critically Ill Patients with Cardiogenic Shock. Medical Science Monitor, 25, 7005-7015. [Google Scholar] [CrossRef] 
[17] Chen, T. and Guestrin, C. (2016) XGBoost: A Scalable Tree Boosting System. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Min-ing, Association for Computing Machinery, San Francisco, 785-794. [Google Scholar] [CrossRef] 

为你推荐