RESUMEN
Background: The long-term prognosis of COVID-19 survivors remains poorly understood. It is evidenced that the lung is the main damaged organ in COVID-19 survivors, most notably in impairment of pulmonary diffusion function. Hence, we conducted a meta-analysis of the potential risk factors for impaired diffusing capacity for carbon monoxide (DLCO) in convalescent COVID-19 patients. Methods: We performed a systematic search of PubMed, Web of Science, Embase, and Ovid databases for relevant studies from inception until January 7, 2022, limited to papers involving human subjects. Studies were reviewed for methodological quality. Fix-effects and random-effects models were used to pool results. Heterogeneity was assessed using I2. The publication bias was assessed using the Egger's test. PROSPERO registration: CRD42021265377. Findings: A total of eighteen qualified articles were identified and included in the systematic review, and twelve studies were included in the meta-analysis. Our results showed that female (OR: 4.011; 95% CI: 2.928-5.495), altered chest computerized tomography (CT) (OR: 3.002; 95% CI: 1.319-6.835), age (OR: 1.018; 95% CI: 1.007-1.030), higher D-dimer levels (OR: 1.012; 95% CI: 1.001-1.023) and urea nitrogen (OR: 1.004;95% CI: 1.002-1.007) were identified as risk factors for impaired DLCO. Interpretation: Pulmonary diffusion capacity was the most common impaired lung function in recovered patients with COVID-19. Several risk factors, such as female, altered chest CT, older age, higher D-dimer levels and urea nitrogen are associated with impairment of DLCO. Raising awareness and implementing interventions for possible modifiable risk factors may be valuable for pulmonary rehabilitation. Funding: This work was financially supported by Emergency Key Program of Guangzhou Laboratory (EKPG21-29, EKPG21-31), Incubation Program of National Science Foundation for Distinguished Young Scholars by Guangzhou Medical University (GMU2020-207).
RESUMEN
The efficient diagnosis of COVID-19 plays a key role in preventing the spread of this disease. The computer-aided diagnosis with deep learning methods can perform automatic detection of COVID-19 using CT scans. However, large scale annotation of CT scans is impossible because of limited time and heavy burden on the healthcare system. To meet the challenge, we propose a weakly-supervised deep active learning framework called COVID-AL to diagnose COVID-19 with CT scans and patient-level labels. The COVID-AL consists of the lung region segmentation with a 2D U-Net and the diagnosis of COVID-19 with a novel hybrid active learning strategy, which simultaneously considers sample diversity and predicted loss. With a tailor-designed 3D residual network, the proposed COVID-AL can diagnose COVID-19 efficiently and it is validated on a large CT scan dataset collected from the CC-CCII. The experimental results demonstrate that the proposed COVID-AL outperforms the state-of-the-art active learning approaches in the diagnosis of COVID-19. With only 30% of the labeled data, the COVID-AL achieves over 95% accuracy of the deep learning method using the whole dataset. The qualitative and quantitative analysis proves the effectiveness and efficiency of the proposed COVID-AL framework.