ABSTRACT
BACKGROUND AND AIM: During the COVID period, reduced exposure to common viral infections is suggested to produce an immune debt in children leading to severe infections with complications. We planned to compare the clinical outcomes of non-covid viral respiratory tract infections (RTI) in children between the pre-COVID and COVID era. METHOD(S): Data from medical records of children admitted with RTI in pre-COVID (2018 - 2019) & COVID era (2021) were analyzed. Patient demographics, virology profile & outcomes were compared. Primary objective was to compare the need for invasive ventilation between the two groups and the secondary objective was to compare the length of ICU stay. RESULT(S): Total number of children admitted with RTI needing oxygen during pre-COVID and COVID era were 140 & 70 respectively. Out of this, 116 and 49 were virology positive. RSV was the commonest virus in both groups. During the pre-COVID period,12 out of 116 children (10.3%) needed invasive ventilation and in the COVID era, 7 out of 49 (14%) were ventilated (Relative risk: 1.38, 95% CI: 0.57 - 3.2). 9 out of 116 children (7%) in the pre- COVID period & 10 out of 49 children (18%) in the COVID era needed prolonged ICU stay (more than 14 days) (Relative risk: 2.63, 95% CI: 1.13 - 6.07). CONCLUSION(S): Children with viral respiratory infection in the COVID era required prolonged ICU stay compared to children in the pre-COVID period.
ABSTRACT
The ongoing COVID-19 pandemic is bringing an “infodemic” on social media. Simultaneously, the huge volume of misinformation (such as rumors, fake news, spam posts, etc.) is scattered in every corner of people’s social life. Traditional misinformation detection methods typically focus on centralized offline processing, that is, they process pandemic-related social data by deploying the model in a single local server. However, such processing incurs extremely long latency when detecting social misinformation related to COVID-19, and cannot handle large-scale social misinformation. In this paper, we propose COS2, a distributed and scalable system that supports large-scale COVID-19-related social misinformation detection. COS2 is able to automatically deploy many groups to distribute deep learning models in scalable cloud servers, process large-scale COVID-19-related social data in various groups, and efficiently detect COVID-19-related tweets with low latency. © 2022, Springer Nature Switzerland AG.