ABSTRACT
ABSTRACT Introduction: COVID-19 disease presentation is heterogeneous, from asymptomatic up to severe life-threatening forms. Getting further insights into patients with specific diseases is of particular interest. We aimed to identify profiles of hematology patients hospitalized with COVID-19 that would be associated with survival and to assess the differences between cohorts Methods: A binational cohort of 263 patients with COVID-19 and hematological disease was studied in Paris, France and São Paulo, Brazil. Patient profiles were based on age, comorbidities, biological measurements, COVID-19 symptoms and hematological disease characteristics. A semi-supervised learning method with a survival endpoint was first used, following which, a classifier was identified to allow the classification of patients using only baseline information Main results: Two profiles of patients were identified, one being young patients with few comorbidities and low C-reactive protein (CRP), D-dimers, lactate dehydrogenase (LDH) and creatinine levels, and the other, older patients, with several comorbidities and high levels of the 4 biology markers. The profiles were strongly associated with survival (p < 0.0001), even after adjusting for age (p = 0.0002). The 30-day survival rate was 77.1% in the first profiles, versus 46.7% in the second. The Brazilian analysis emphasized the importance of age, while the French focused on the comorbidities Conclusion: This analysis showed the importance of CRP, LHD and creatinine in the COVID-19 presentation and prognosis, whatever the geographic origin of the patients.
Subject(s)
COVID-19 , Prognosis , Clinical Laboratory Techniques , Hematologic DiseasesABSTRACT
INTRODUCTION: COVID-19 disease presentation is heterogeneous, from asymptomatic up to severe life-threatening forms. Getting further insights into patients with specific diseases is of particular interest. We aimed to identify profiles of hematology patients hospitalized with COVID-19 that would be associated with survival and to assess the differences between cohorts METHODS: A binational cohort of 263 patients with COVID-19 and hematological disease was studied in Paris, France and São Paulo, Brazil. Patient profiles were based on age, comorbidities, biological measurements, COVID-19 symptoms and hematological disease characteristics. A semi-supervised learning method with a survival endpoint was first used, following which, a classifier was identified to allow the classification of patients using only baseline information MAIN RESULTS: Two profiles of patients were identified, one being young patients with few comorbidities and low C-reactive protein (CRP), D-dimers, lactate dehydrogenase (LDH) and creatinine levels, and the other, older patients, with several comorbidities and high levels of the 4 biology markers. The profiles were strongly associated with survival (p < 0.0001), even after adjusting for age (p = 0.0002). The 30-day survival rate was 77.1% in the first profiles, versus 46.7% in the second. The Brazilian analysis emphasized the importance of age, while the French focused on the comorbidities CONCLUSION: This analysis showed the importance of CRP, LHD and creatinine in the COVID-19 presentation and prognosis, whatever the geographic origin of the patients.
ABSTRACT
Automatic recognition of visual objects using a deep learning approach has been successfully applied to multiple areas. However, deep learning techniques require a large amount of labeled data, which is usually expensive to obtain. An alternative is to use semi-supervised models, such as co-training, where multiple complementary views are combined using a small amount of labeled data. A simple way to associate views to visual objects is through the application of a degree of rotation or a type of filter. In this work, we propose a co-training model for visual object recognition using deep neural networks by adding layers of self-supervised neural networks as intermediate inputs to the views, where the views are diversified through the cross-entropy regularization of their outputs. Since the model merges the concepts of co-training and self-supervised learning by considering the differentiation of outputs, we called it Differential Self-Supervised Co-Training (DSSCo-Training). This paper presents some experiments using the DSSCo-Training model to well-known image datasets such as MNIST, CIFAR-100, and SVHN. The results indicate that the proposed model is competitive with the state-of-art models and shows an average relative improvement of 5% in accuracy for several datasets, despite its greater simplicity with respect to more recent approaches.