No abstract

Introduction: To retrospectively analyze epidemiological, clinical and hematological characteristics of COVID-19 patients. Methods: The demographic, symptoms, and physiological parameters of 88 patients were collected and analyzed. The performance of complete blood count (CBC) indexes for monitoring and predicting the severity of COVID-19 in patients was evaluated by analyzing and comparing CBC results among different COVID-19 patient groups. Results: White blood cells (WBCs), the neutrophil percentage (Neu%), absolute neutrophil count (Neu#), and neutrophil-to-lymphocyte ratio (NLR) were significantly higher in the critical group than in the other three groups (P < .05), while the lymphocyte percentage (Lym%), monocyte percentage (Mon%), lymphocyte count (Lym#), and lymphocyte-to-monocyte ratio (LMR) were significantly lower in the critical group than in the other three groups (P < .05). WBCs, the Neu%, Neu#, NLR, and neutrophil-to-monocyte ratio (NMR) were significantly higher in the severe group than in the mild and moderate groups (P < .05), while the Lym% was significantly lower in the severe group than in the mild and moderate groups (P < .05). The Mon%, Lym#, and LMR were significantly lower in the severe group than in the moderate group (P < .05). Using receiver operating characteristic (ROC) curve analysis to differentiate severe and nonsevere patients, the areas under the curve (AUCs) for the NLR, Neu%, and Lym% were 0.733, 0.732, and 0.730, respectively. When differentiating critical patients from noncritical patients, the AUCs for the NLR, Neu%, and Lym% were 0.832, 0.831, and 0.831. Conclusions: The NLR is valuable for differentiating and predicting patients who will become critical within 4 weeks after the onset of COVID-19.

In this paper, we study physical adversarial attacks on object detectors in the wild. Prior arts on this matter mostly craft instance-dependent perturbations only for rigid and planar objects. To this end, we propose to learn an adversarial pattern to effectively attack all instances belonging to the same object category (e.g., person, car), referred to as Universal Physical Camouflage Attack (UPC). Concretely, UPC crafts camouflage by jointly fooling the region proposal network, as well as misleading the classifier and the regressor to output errors. In order to make UPC effective for articulated non-rigid or non-planar objects, we introduce a set of transformations for the generated camouflage patterns to mimic their deformable properties. We additionally impose optimization constraint to make generated patterns look natural for human observers. To fairly evaluate the effectiveness of different physical-world attacks on object detectors, we present the first standardized virtual database, AttackScenes, which simulates the real 3D world in a controllable and reproducible environment. Extensive experiments suggest the superiority of our proposed UPC compared with existing physical adversarial attackers not only in virtual environments (AttackScenes), but also in real-world physical environments. Codes, models, and demos are publicly available at https://mesunhlf. github.io/index_physical.html.

No abstract