机器学习预测ICU死亡率和心房的停留时间:MIMIC-IV/MIMIC-III研究
Victoria Nguyen1, Rahul Mittal1
1Department of Health Informatics, Rutgers University, Piscataway, NJ 08854, USA.
Healthcare (Basel, Switzerland)
|February 13, 2026
概括
机器学习模型可以使用早期数据预测心房动 (AF) 患者的重症监护室 (ICU) 死亡率. 然而,根据目前的数据,预测ICU停留时间 (LOS) 仍然具有挑战性.
科学领域:
- 关键护理医学 关键护理医学
- 心脏病学 心脏病学
- 数据科学与机器学习
背景情况:
- 前庭动 (AF) 在重症监护室 (ICU) 很普遍,与增加死亡率和长时间停留 (LOS) 等不良结果相关.
- 现有的AF风险评分在重病人群中缺乏适用性.
- 这项研究解决了针对FI的ICU患者量身定制的预测模型的需求.
研究的目的:
- 为了描述患有心房动 (AF) 的重症监护病房 (ICU) 患者.
- 开发和验证机器学习 (ML) 模型,用于预测AF患者的ICU死亡率和停留时间 (LOS).
- 使用可解释的ML方法识别与ICU死亡率和LOS相关的早期临床因素.
主要方法:
- 从MIMIC-IV (n=20,058) 采用了FI的成年ICU患者,用于模型开发和MIMIC-III (n=11,475) 用于时间外部验证.
- 包括人口统计学,入院特征,生命体征,实验室,血管活性支持,以及前24小时内AF药物作为预测因素.
- 评估了多种分类和回归算法,评估死亡率的AUC/AP和LOS的MAE/RMSE/R2的歧视,使用SHAP进行解释性.
主要成果:
- 在XGBoost模型中,在时间验证 (AUC=0.743) 上显示了ICU死亡率预测的保留歧视.
- 优化概率值提高了在低患病率场景中对死亡率预测的灵敏度.
- 机器学习模型解释了ICU LOS预测的最小差异 (R2=0.038),表明了早期数据的限制.
结论:
- 机器学习模型,特别是XGBoost,显示了利用早期临床数据预测AF患者的ICU死亡率的潜力.
- 仅基于前24小时的数据来预测ICU停留时间 (LOS) 是一个挑战,这表明后来的临床和操作因素的影响.
- 可解释方法确定了不良结果的关键早期预测因素,包括尿量减少,功能障碍,代谢障碍,低氧化,血管压缩剂使用和晚年.
相关概念视频
03:03Amyloid Fibrils
12.1K
Amyloid fibrils are aggregates of misfolded proteins. Under most circumstances, misfolded proteins are either refolded by chaperone proteins or degraded by the proteasome. However, in the case of a mutation or a disease, these proteins can accumulate to form large clusters and often further assemble to form elongated fibers, called fibrils.
Amyloid deposits were observed as early as 1639 in the liver and the spleen. In 1854, Rudolph Virchow performed iodine staining,...
Amyloid deposits were observed as early as 1639 in the liver and the spleen. In 1854, Rudolph Virchow performed iodine staining,...
12.1K
03:03Amyloid Fibrils
6.5K
6.5K
01:43Electron Transport Chain: Complex III and IV
9.3K
During the electron transport chain, electrons from NADH and FADH2 are first transferred to complexes I and II, respectively. These two complexes then transfer the electrons to ubiquinol, which carries them further to complex III. Complex III passes the electrons across the intermembrane space to Cyt c, which carries them further to complex IV. Complex IV donates electrons to oxygen and reduces it to water. As electrons pass through complexes I, III, and IV, the energy released aids the pumping...
9.3K
02:27Predicting Molecular Geometry
46.1K
VSEPR Theory for Determination of Electron Pair Geometries
46.1K
01:21Imaging Studies for Cardiovascular System IV: CMRI
419
Cardiovascular magnetic resonance imaging, or CMRI, is a non-invasive diagnostic test that employs a magnetic field and radiofrequency waves to create precise images of the heart and arteries. It provides comprehensive information about cardiac anatomy, function, perfusion, and tissue characterization without ionizing radiation.IndicationsCMRI diagnoses various heart conditions, including tissue damage from heart attacks, ischemic heart disease, myocarditis, aortic issues (tears, aneurysms,...
419
01:11Fibril-associated Collagen
3.4K
Fibril-associated collagens are a type of collagens present in the extracellular matrix with interrupted triple helices or FACIT (Fibril-associated collagens interrupted triple-helices). FACIT help connect and attach the collagen fibrils with each other as well as with other proteins of the extracellular matrix.
For example, the type II collagen fibrils in cartilage have covalently bound type IX fibril-associated collagens at regular intervals. Other types of fibril-associated collagens are...
For example, the type II collagen fibrils in cartilage have covalently bound type IX fibril-associated collagens at regular intervals. Other types of fibril-associated collagens are...
3.4K