人工智能驱动的胸部计算机断层扫描分析揭示了普通血液透析患者中COVID-19死亡率的预后见解。

Artificial intelligence-powered chest computed tomography analysis unveils prognostic insights for COVID-19 mortality among prevalent hemodialysis patients.

作者信息

Kim Eunji, Yoon Soo-Jin, Yu Sungbong, Ko Eunsil, Shon Kyungjun, Yoon Jooyeon, Kee Youn Kyung, Kim Do Hyoung, Cho AJin, Park Hayne Cho, Lee Young-Ki

机构信息

Department of Internal Medicine, Kangnam Sacred Heart Hospital, Hallym University College of Medicine, Seoul, Republic of Korea.

Hallym University Kidney Research Institute, Seoul, Republic of Korea.

出版信息

Kidney Res Clin Pract. 2024 Sep 26. doi: 10.23876/j.krcp.24.079.

DOI:10.23876/j.krcp.24.079

PMID:39384348

Abstract

BACKGROUND

Coronavirus disease 2019 (COVID-19) has led to severe pneumonia and mortality worldwide, however, clinical outcomes in end-stage renal disease patients remain unclear. This study evaluates the prognostic value of chest computed tomography (CT) findings in predicting COVID-19-related outcomes in prevalent hemodialysis patients.

METHODS

We retrospectively analyzed 326 prevalent hemodialysis patients diagnosed with COVID-19 who underwent chest CT scans. Characteristics assessed included pleural effusion, lung involvement volume, nodular consolidation, patchy infiltration, and ground-glass opacity. Artificial intelligence (AI)-assisted CT analysis quantified lung involvement. The primary endpoint was in-hospital mortality. Clinical data were collected, and logistic regression analysis assessed the association between CT findings and mortality.

RESULTS

The mean age of the patients was 66.7 ± 12.6 years, 61.0% were male, and 58.6% were diabetic. Chest CT showed that 18.1% had lung involvement >10%, 32.5% had pleural effusion, 68.7% had nodular consolidation, 57.1% had patchy infiltration, and 58.0% had ground-glass opacity. Seventy patients (21.5%) died. Multivariate logistic regression analysis identified lung involvement >2.7% (odds ratio [OR], 16.70; 95% confidence interval [CI], 4.35-65.63), pleural effusion (OR, 3.28; 95% CI, 1.15-9.35), nodular consolidation (OR, 4.08; 95% CI, 1.12-14.82), and patchy infiltration (OR, 3.75; 95% CI, 1.17-12.03) as significant mortality risk factors.

CONCLUSION

Chest CT findings, including lung involvement >2.7% and the presence of pleural effusion, nodular consolidation, and patchy infiltrates, significantly indicated mortality in COVID-19 pneumonia among prevalent hemodialysis patients. AI-assisted CT analysis proved useful in assessing lung involvement extent, showing that even minimal lung involvement can be associated with increased mortality.

摘要

背景

2019冠状病毒病（COVID-19）已在全球范围内导致严重肺炎和死亡，然而，终末期肾病患者的临床结局仍不明确。本研究评估胸部计算机断层扫描（CT）结果在预测维持性血液透析患者COVID-19相关结局中的预后价值。

方法

我们回顾性分析了326例确诊为COVID-19并接受胸部CT扫描的维持性血液透析患者。评估的特征包括胸腔积液、肺受累体积、结节状实变、斑片状浸润和磨玻璃影。人工智能（AI）辅助CT分析对肺受累情况进行量化。主要终点是住院死亡率。收集临床数据，采用逻辑回归分析评估CT结果与死亡率之间的关联。

结果

患者的平均年龄为66.7±12.6岁，男性占61.0%，糖尿病患者占58.6%。胸部CT显示，18.1%的患者肺受累>10%，32.5%的患者有胸腔积液，68.7%的患者有结节状实变，57.1%的患者有斑片状浸润，58.0%的患者有磨玻璃影。70例患者（21.5%）死亡。多因素逻辑回归分析确定肺受累>2.7%（比值比[OR]，16.70；95%置信区间[CI]，4.35-65.63）、胸腔积液（OR，3.28；95%CI，1.15-9.35）、结节状实变（OR，4.08；95%CI，1.12-14.82）和斑片状浸润（OR，3.75；95%CI，1.17-12.03）为显著的死亡风险因素。