Venous Thrombo-Embolism in an Outpatient Vascular Ultrasound Laboratory All-Comers Cohort with Recent COVID-19 Infection

Introduction: Inflammatory mechanisms triggered and supported by SARS-CoV-2 infection can increase venous thromb-oembolism (VTE) risk. Aim: The aim of the present comparative study was to report on the incidence of VTE in a prospective consecutive series of COVID-19 negative outpatients referred to our vascular ultrasound laboratory for suspected VTE with (COVID+) or without (COVID–) recent SARS-CoV-2 infection during the COVID-19 pandemic. Methods: All patients included in the present study were assessed by duplex ultrasound for the detection of VTE. The following data were collected for each patient: time from first negative swab after COVID-19 infection and VTE diagnosis in COVID+ patients;administration of heparin prophylaxis during COVID-19 infection;presence of venous risk factors (previous VTE episode, chronic venous insufficiency, thrombophilia, recent surgery with prolonged immobilisation, history of malignancy, need for high dose steroid therapy, and dehydration during infection). Rate of VTE was detected and compared between the two groups of COVID+ or COVID– patients by chi square test for categorical data. The presence of risk factors for VTE were analysed as related to VTE occurrence in both groups. Results: From 1 February 2021 to 31 March 2021, 34 patients were included in the study. Among them eight had previous SARS-CoV-2 infection and were negative at the time of investigation. In COVID+ patients, time from first negative swab after COVID-19 infection and VTE diagnosis ranged between 3 and 50 days (mean 17 ± 14.39 days), and 12.5% (n = 1/8) had heparin prophylaxis during infection. Risk factors for VTE were detected in all COVID+ patients and 80.7% (n = 21/26) of COVID– patients. Rate of VTE was 87.5% (n = 7/8) in COVID+ patients and 11.5% (n = 3/26) in COVID– patients (odds ratio 53.66, 95% confidence interval 4.79 – 601.23;p <.001). In the COVID+ population only one patient receiving heparin prophylaxis during infection did not present with VTE. One COVID+ patient presented with both arterial and venous popliteal thrombosis. By comparing directly COVID+ patients with no heparin prophylaxis and venous risk factors (n = 7) to COVID– patients with venous risk factors (n = 21), VTE risk was strongly associated with the presence of previous SARS-CoV-2 infection without proper heparin prophylactic administration (p <.001). Conclusion: In this preliminary series presence of risk factors for VTE and recent SARS-CoV-2 infection with no heparin prophylaxis is strongly associated with VTE occurrence. COVID-19 outpatients should be treated by prophylactic heparin whenever VTE risk factors are detected and duplex ultrasound cannot be performed to exclude a VTE episode, so that physicians treating outpatients should be aware of the VTE risk in those patients. Both arterial and venous conditions prone to thrombosis should be fully assessed in patients when diagnosing a new SARS-CoV-2 infection.

Submillisievert CT chest for COVID-19 patients in a rural hospital with limited resources

Background: This is a secondary analysis of prospectively acquired data approved by the hospital institutional board committee. We performed a retrospective chart review of 463 patients who underwent a CT Chest for suspected COVID-19 infection between April 1st, 2020, and March 31st, 2021. Patients were grouped based on the CT chest obtained protocol: ultra-low dose or full dose. The likelihood of suspicion of COVID-19 infection was classified on a Likert scale based on the probability of pulmonary involvement. For each group, the sensitivity and specificity of CT were compared to nasopharyngeal swab as standard of reference. The median dose length product and duration of apnea were compared between both groups using two-tailed Mann–Whitney U test. The aim of this study is to share our experience of reducing radiation dose in COVID-19 patients by using an ultra-low dose CT chest protocol on a 16 row multidetector CT scan in a hospital with limited resources. Results: Two hundred sixty-nine patients underwent a full dose CT and 194 patients an ultra-low dose CT. In the former group, the median dose length product was 341.11 mGy*cm [Interquartile range (IQR), 239.1–443.2] and the median duration of apnea was 13.29 s [IQR, 10.85–15.73]. In the latter group, the median dose length product was 30.8 mGy*cm [IQR, 28.9–32.7] and median duration of apnea was 8.27 s [IQR, 7.69–8.85]. The sensitivity of the ultra-low dose CT was 91.2% and that of the full dose was 94%. Conclusion: A 90% reduction in estimated dose and 38% reduction in apnea duration could be achieved using an ultra-low dose CT chest protocol on a 16-row MDCT without significant loss in the sensitivity of CT to detect COVID-related parenchymal involvement.

Did I See It Before? Detecting Previously-Checked Claims over Twitter

With the proliferation of fake news in the last few years, especially during the COVID-19 period, combating the spread of misinformation has become an urgent need. Although automated fact-checking systems were proposed recently, they leave much to be desired in terms of accuracy and explainability. Therefore, involving humans during verification could make the process much easier and more reliable. In this work, we propose an automated approach to detect claims that have been already manually-verified by professional fact-checkers. Our proposed approach uses recent powerful BERT variants as point-wise rerankers. Additionally, we study the impact of using different fields of the verified claim during training and inference phases. Experimental results show that our proposed pipeline outperforms the state-of-the-art approaches on two English and one Arabic datasets.

Explaining COVID-19 and Thoracic Pathology Model Predictions by Identifying Informative Input Features

Neural networks have demonstrated remarkable performance in classification and regression tasks on chest X-rays. In order to establish trust in the clinical routine, the networks’ prediction mechanism needs to be interpretable. One principal approach to interpretation is feature attribution. Feature attribution methods identify the importance of input features for the output prediction. Building on Information Bottleneck Attribution (IBA) method, for each prediction we identify the chest X-ray regions that have high mutual information with the network’s output. Original IBA identifies input regions that have sufficient predictive information. We propose Inverse IBA to identify all informative regions. Thus all predictive cues for pathologies are highlighted on the X-rays, a desirable property for chest X-ray diagnosis. Moreover, we propose Regression IBA for explaining regression models. Using Regression IBA we observe that a model trained on cumulative severity score labels implicitly learns the severity of different X-ray regions. Finally, we propose Multi-layer IBA to generate higher resolution and more detailed attribution/saliency maps. We evaluate our methods using both human-centric (ground-truth-based) interpretability metrics, and human-agnostic feature importance metrics on NIH Chest X-ray8 and BrixIA datasets. The code (https://github.com/CAMP-eXplain-AI/CheXplain-IBA ) is publicly available. © 2021, Springer Nature Switzerland AG.

Overview of the CLEF–2021 CheckThat! Lab on Detecting Check-Worthy Claims, Previously Fact-Checked Claims, and Fake News

We describe the fourth edition of the CheckThat! Lab, part of the 2021 Conference and Labs of the Evaluation Forum (CLEF). The lab evaluates technology supporting tasks related to factuality, and covers Arabic, Bulgarian, English, Spanish, and Turkish. Task 1 asks to predict which posts in a Twitter stream are worth fact-checking, focusing on COVID-19 and politics (in all five languages). Task 2 asks to determine whether a claim in a tweet can be verified using a set of previously fact-checked claims (in Arabic and English). Task 3 asks to predict the veracity of a news article and its topical domain (in English). The evaluation is based on mean average precision or precision at rank k for the ranking tasks, and macro-F1 for the classification tasks. This was the most popular CLEF-2021 lab in terms of team registrations: 132 teams. Nearly one-third of them participated: 15, 5, and 25 teams submitted official runs for tasks 1, 2, and 3, respectively. © 2021, Springer Nature Switzerland AG.

Sleep apnea severity and COVID-19 hospitalization outcomes: Interim analysis

Introduction: Millions worldwide have been infected with COVID- 19. Risk factors for poor outcomes include older age, obesity, cardiovascular disease, diabetes, COPD, and minority ethnicity. The impact of sleep apnea (OSA) on COVID-19 illness remains a topic of ongoing research. The association seems plausible as OSA shares common risk factors with COVID-19 comorbidities. The aim of our study is to investigate if the severity of OSA correlates with COVID-19 hospitalization outcomes. Methods: The study is an observational retrospective electronic medical chart review of patients admitted to the Yale Health system for COVID-19 illness, who have an ICD medical history diagnosis of sleep apnea. Subjects with an available sleep study were grouped based on AHI severity. Composite outcome was determined by mortality or critical illness (ICU admission, mechanical ventilation, High-Flow nasal cannula or noninvasive ventilation). Interim analysis was conducted with data from 256 patients. Logistic regression was performed to calculate OR associated with the primary outcome comparing mild versus moderate-severe OSA groups. Results: The sample of 256 patients included 50% females, with a median age of 64 (IQR,55-73) and BMI of 36 (IQR,30-41). Race distribution composed of 45% Whites and 40% African-Americans with 18% identifying as Hispanic. Overall mortality rate was 16%. Median length of stay (LOS) was 9 days (IQR,5 -15). 155 patients had a recorded AHI with a median of 26/hour (IQR,11-51) grouped into mild (34%), moderate (20%) and severe (45%) disease. Severe sleep apnea had the highest mortality rate of 19% and median LOS of 10 days (IQR,5- 16) compared to moderate (6% mortality,9 days (IQR,6-14)) and mild (17% mortality,6 days (IQR,4-11)) disease;differences were not statistically significant. Univariate logistic regression analysis demonstrated no significant difference in the composite outcome for mild versus moderate-severe OSA groups (OR=1.2;95% CI:0.58-2.32). Conclusion: Severe OSA appeared to have a trend towards an association with higher mortality versus moderate but not mild disease. Comparing moderate-severe disease to a reference group of mild disease did not demonstrate a significant difference in our composite outcome of death or critical illness. Additional subject recruitment and re-analysis are needed to confirm the findings of this interim analysis.

Overview of the CLEF-2021 CheckThat! Lab Task 2 on detecting previously fact-checked claims in tweets and political debates

We describe the fourth edition of the CheckThat! Lab, part of the 2021 Conference and Labs of the Evaluation Forum (CLEF). The lab evaluates technology supporting three tasks related to factuality, and it covers Arabic, Bulgarian, English, Spanish, and Turkish. Here, we present the task 2, which asks to detect previously fact-checked claims (in two languages). A total of four teams participated in this task, submitted a total of sixteen runs, and most submissions managed to achieve sizable improvements over the baselines using transformer based models such as BERT, RoBERTa. In this paper, we describe the process of data collection and the task setup, including the evaluation measures used, and we give a brief overview of the participating systems. Last but not least, we release to the research community all datasets from the lab as well as the evaluation scripts, which should enable further research in detecting previously fact-checked claims. © 2021 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).