论文已发表
注册即可获取德孚的最新动态
IF 收录期刊
基于LASSO的机器学习算法预测脓毒症并发PICS
Authors Hui K, Hong C, Xiong Y, Xia J, Huang W , Xia A, Xu S, Chen Y, Zhang Z, Chen H
Received 10 April 2024
Accepted for publication 20 June 2024
Published 1 July 2024 Volume 2024:17 Pages 2701—2710
DOI https://doi.org/10.2147/IDR.S464906
Checked for plagiarism Yes
Review by Single anonymous peer review
Peer reviewer comments 3
Editor who approved publication: Prof. Dr. Héctor Mora-Montes
Kangping Hui,1,* Chengying Hong,2,* Yihan Xiong,3 Jinquan Xia,4 Wei Huang,5 Andi Xia,2 Shunyao Xu,2 Yuting Chen,2 Zhongwei Zhang,6 Huaisheng Chen2
1The Second Clinical Medical College, Jinan University, Shenzhen, Guangdong Province, People’s Republic of China; 2Department of Critical Care Medicine, Shenzhen People’s Hospital, Second Clinical Medical College of Jinan University, First Affiliated Hospital of Southern University of Science and Technology, Shenzhen, Guangdong Province, People’s Republic of China; 3Neurology Department, Shenzhen People’s Hospital, Second Clinical Medical College of Jinan University, First Affiliated Hospital of Southern University of Science and Technology, Shenzhen, Guangdong Province, People’s Republic of China; 4Department of Clinical Medical Research Center, the Second Clinical Medical College, Jinan University (Shenzhen People’s Hospital), the First Affiliated Hospital of Southern University of Science and Technology, Shenzhen, Guangdong Province, People’s Republic of China; 5Department of Clinical Microbiology, Shenzhen People’s Hospital, Second Clinical Medical College of Jinan University, First Affiliated Hospital of Southern University of Science and Technology, Shenzhen, Guangdong Province, People’s Republic of China; 6Department of Critical Care Medicine, West China Hospital, Sichuan University, Chengdu, Sichuan Province, People’s Republic of China
*These authors contributed equally to this work
Correspondence: Huaisheng Chen, Department of Critical Care Medicine, Shenzhen People’s Hospital, Second Clinical Medical College of Jinan University, First Affiliated Hospital of Southern University of Science and Technology, Shenzhen, Guangdong Province, People’s Republic of China, Email sunshinic@hotmail.com Zhongwei Zhang, Department of Critical Care Medicine, West China Hospital, Sichuan University, Chengdu, Sichuan Province, People’s Republic of China, Email 2422708196@qq.com
Introduction: This study aims to establish a comprehensive, multi-level approach for tackling tropical diseases by proactively anticipating and managing Persistent Inflammation, Immunosuppression, and Catabolism Syndrome (PICS) within the initial 14 days of Intensive Care Unit (ICU) admission. The primary objective is to amalgamate a diverse array of indicators and pathogenic microbial data to pinpoint pivotal predictive variables, enabling effective intervention specifically tailored to the context of tropical diseases.
Methods: A focused analysis was conducted on 1733 patients admitted to the ICU between December 2016 and July 2019. Utilizing the Least Absolute Shrinkage and Selection Operator (LASSO) regression, disease severity and laboratory indices were scrutinized. The identified variables served as the foundation for constructing a predictive model designed to forecast the occurrence of PICS.
Results: Among the subjects, 13.79% met the diagnostic criteria for PICS, correlating with a mortality rate of 38.08%. Key variables, including red-cell distribution width coefficient of variation (RDW-CV), hemofiltration (HF), mechanical ventilation (MV), Norepinephrine (NE), lactic acidosis, and multiple-drug resistant bacteria (MDR) infection, were identified through LASSO regression. The resulting predictive model exhibited a robust performance with an Area Under the Curve (AUC) of 0.828, an accuracy of 0.862, and a specificity of 0.977. Subsequent validation in an independent cohort yielded an AUC of 0.848.
Discussion: The acquisition of RDW-CV, HF requirement, MV requirement, NE requirement, lactic acidosis, and MDR upon ICU admission emerges as a pivotal factor for prognosticating PICS onset in the context of tropical diseases. This study highlights the potential for significant improvements in clinical outcomes through the implementation of timely and targeted interventions tailored specifically to the challenges posed by tropical diseases.
Keywords: sepsis, persistent inflammation immunosuppression catabolism syndrome, LASSO regression, mortality, predictive model