Background: A COVID-19 outbreak started in Wuhan, China, last December and now has become a global pandemic. The clinical information in caring of critically ill patients with COVID-19 needs to be shared timely, especially under the situations that there is still a largely ongoing spread of COVID-19 in many countries. Methods: A multicenter prospective observational study investigated all the COVID-19 patients received in 19 ICUs of 16 hospitals in Wuhan, China, over 24 h between 8 AM February 2h and 8 AM February 27, 2020. The demographic information, clinical characteristics, vital signs, complications, laboratory values, and clinical managements of the patients were studied.Results: A total of 226 patients were included. Their median (interquartile range, IQR) age was 64 (57-70) years, and 139 (61.5%) patients were male. The duration from the date of ICU admission to the study date was 11 (5-17) days, and the duration from onset of symptoms to the study date was 31 (24-36) days. Among all the patients, 155 (68.6%) had at least one coexisting disease, and their sequential organ failure assessment score was 4 (2-8). Organ function damages were found in most of the patients: ARDS in 161 (71.2%) patients, septic shock in 34 (15.0%) patients, acute kidney injury occurred in 57 (25.2%) patients, cardiac injury in 61 (27.0%) patients, and lymphocytopenia in 160 (70.8%) patients. Of all the studied patients, 85 (37.6%) received invasive mechanical ventilation, including 14 (6.2%) treated with extracorporeal membrane oxygenation (ECMO) at the same time, 20 (8.8%) received noninvasive mechanical ventilation, and 24 (10.6%) received continuous renal replacement therapy. By April 9, 2020, 87 (38.5%) patients were deceased and 15 (6.7%) were still in the hospital.(Continued on next page)

These authors contributed equally to this work. ⇤ Corresponding authors.The sudden outbreak of novel coronavirus 2019 (COVID-19) increased the diagnostic burden of radiologists. In the time of an epidemic crisis, we hoped artificial intelligence (AI) to help reduce physician workload in regions with the outbreak, and improve the diagnosis accuracy for physicians before they could acquire enough experience with the new disease. Here, we present our experience in building and deploying an AI system that automatically analyzes CT images to detect COVID-19 pneumonia features. Different from conventional medical AI, we were dealing with an epidemic crisis. Working in an interdisciplinary team of over 30 people with medical and / or AI background, geographically distributed in Beijing and Wuhan, we were able to overcome a series of challenges in this particular situation and deploy the system in four weeks. Using 1,136 training cases (723 positives for COVID-19) from five hospitals, we were able to achieve a sensitivity of 0.974 and specificity of 0.922 on the test dataset, which included a variety of pulmonary diseases. Besides, the system automatically highlighted all lesion regions for faster examination. As of today, we have deployed the system 2 All rights reserved. No reuse allowed without permission.

No abstract

Deep learning (DL) systems are increasingly applied to safetycritical domains such as autonomous driving cars. It is of significant importance to ensure the reliability and robustness of DL systems. Existing testing methodologies always fail to include rare inputs in the testing dataset and exhibit low neuron coverage.In this paper, we propose DLFuzz, the first differential fuzzing testing framework to guide DL systems exposing incorrect behaviors. DLFuzz keeps minutely mutating the input to maximize the neuron coverage and the prediction difference between the original input and the mutated input, without manual labeling effort or cross-referencing oracles from other DL systems with the same functionality. We present empirical evaluations on two well-known datasets to demonstrate its efficiency. Compared with DeepXplore, the state-of-the-art DL whitebox testing framework, DLFuzz does not require extra efforts to find similar functional DL systems for cross-referencing check, but could generate 338.59% more adversarial inputs with 89.82% smaller perturbations, averagely obtain 2.86% higher neuron coverage, and save 20.11% time consumption. CCS CONCEPTS• Software and its engineering → Software testing and debugging;

No abstract

No abstract

No abstract

No abstract