Age-stratified analysis of descending aorta diameter in traumatic massive hemorrhage: a machine learning approach

基于机器学习方法的创伤性大出血患者降主动脉直径年龄分层分析:

阅读:1

Abstract

BACKGROUND: Aortic diameter (AoD) changes with age and can decrease in shock states. Accurate AoD assessment is crucial for managing hypovolemic shock and guiding interventions such as resuscitative endovascular balloon occlusion of the aorta. This study hypothesized that clinical factors (eg, initial hemodynamic parameters, trauma severity, and laboratory results) would have a greater impact on the AoD than would age- or anthropometric-related factors in traumatic massive hemorrhage patients. We aimed to identify significant predictors of the descending AoD across two age groups (18-60 years and 61-91 years). METHODS: A retrospective analysis was conducted on 243 massive hemorrhage patients at a level I trauma center. The aorta was automatically segmented in CT images via a deep learning architecture based on a Shallow Attention Network to obtain diaphragm-level AoD values. 152 patients were assigned to the younger group and 91 to the senior group. A random forest model was used to incorporate various clinical factors. RESULTS: In the younger group, age and body surface area were the most important features (root mean square error (RMSE): train, 1.03; test, 2.70). In the senior group, hemoglobin, arterial pH, and heart rate were the most significant indicators (RMSE: train, 1.19; test, 3.95). The importance of age diminished in the senior group, whereas vital signs and laboratory values gained prominence. CONCLUSION: Our findings reveal age-specific differences in factors influencing the AoD during traumatic hemorrhage. The results highlight the limitations of traditional methods for AoD estimation, especially in senior patients in whom dynamic physiological factors may play a major role. These insights can improve the accuracy of AoD assessment and management in hemorrhage patients across different age groups. The findings may contribute to developing an artificial intelligence-derived score that estimates the AoD, incorporating static and dynamic physiological factors. LEVEL OF EVIDENCE: IV, retrospective study having more than one negative criterion.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。