Research progress on prognostic prediction models for patients undergoing extracorporeal membrane oxygenation.
10.3760/cma.j.cn121430-20240715-00598
- Author:
Hanming GAO
1
;
Junyu LU
Author Information
1. Department of Critical Care Medicine, the Second Affiliated Hospital of Guangxi Medical University, Nanning 530007, Guangxi Zhuang Autonomous Region, China. Gao Hanming is working on the Department of Critical Care Medicine, the People's Hospital of Cenxi City, Cenxi 543200, Guangxi Zhuang Autonomous Region, China. Corresponding author: Lu Junyu, Email: junyulu@gxmu.edu.cn.
- Publication Type:English Abstract
- MeSH:
Extracorporeal Membrane Oxygenation/methods*;
Humans;
Prognosis;
Respiratory Distress Syndrome/therapy*;
Machine Learning;
Shock, Cardiogenic/therapy*
- From:
Chinese Critical Care Medicine
2024;36(12):1334-1339
- CountryChina
- Language:Chinese
-
Abstract:
Extracorporeal membrane oxygenation (ECMO), as a critical life support technology, has played a significant role in treating patients with refractory respiratory and circulatory failure. In recent years, with the advancements in medical technology, the scope of application of ECMO has been expanding, especially in the fields of acute respiratory distress syndrome, cardiogenic shock and other important roles. However, its high costs, complex operation, and associated risks of complications remain challenges in clinical practice. At present, an increasing number of studies have focused on developing and validating ECMO prognostic models. Developing precise prognostic prediction models is crucial for optimizing treatment decisions and improving patient survival rates. This article categorizes existing prognostic models for adult ECMO patients based on methodological classification, patient population, and theoretical framework. It highlights the limitations of current models in terms of sample size, multi-center validation, static data analysis, and model applicability. Moreover, it proposes future directions for model development, such as multi-center prospective studies, integration of machine learning and deep learning technologies, and increased focus on long-term outcomes, offering insights for researchers to improve model construction and explore new research directions.
