Weakly Semi-Supervised Detection in Lung Ultrasound Videos (preprint)

Frame-by-frame annotation of bounding boxes by clinical experts is often required to train fully supervised object detection models on medical video data. We propose a method for improving object detection in medical videos through weak supervision from video-level labels. More concretely, we aggregate individual detection predictions into video-level predictions and extend a teacher-student training strategy to provide additional supervision via a video-level loss. We also introduce improvements to the underlying teacher-student framework, including methods to improve the quality of pseudo-labels based on weak supervision and adaptive schemes to optimize knowledge transfer between the student and teacher networks. We apply this approach to the clinically important task of detecting lung consolidations (seen in respiratory infections such as COVID-19 pneumonia) in medical ultrasound videos. Experiments reveal that our framework improves detection accuracy and robustness compared to baseline semi-supervised models, and improves efficiency in data and annotation usage.

REDUCED HEALTH-RELATED QUALITY OF LIFE USING PROMIS IN PATIENTS WITH SYSTEMIC SCLEROSIS: A CROSS-SECTIONAL ANALYSIS FROM THE INTERNATIONAL COVAD-2 E-SURVEY

BackgroundThe second COVID-19 vaccination in autoimmune disease (COVAD-2) study [1] is an international, multicentre, self-reported e-survey designed to evaluate several facets covering COVID-19 infection and vaccination as well as validated patient-reported outcome measures (PROMs) in a variety of autoimmune diseases (AIDs), including systemic sclerosis (SSc). Detailed assessment of the health-related quality of life (HRQOL) and its drivers in patients with SSc is lacking.ObjectivesTo assess physical and mental health in a global cohort of SSc patients in comparison with non-SSc autoimmune inflammatory rheumatic diseases (AIRDs), non-rheumatic AIDs (NRAIDs), and those without AIDs (controls) using Patient-Reported Outcome Measurement Information System (PROMIS) global health data from the COVAD-2 survey.MethodsThe COVAD-2 database was used to extract demographics, AID diagnosis, comorbidities, disease activity, current therapies, and PROMs. PROMIS global physical health (GPH), global mental health (GMH) scores, PROMIS physical function short form-10a (PROMIS PF-10a), pain visual analogue scale (VAS), and PROMIS Fatigue-4a scores were compared between SSc, non-SSc AIRDs, NRAIDs, and controls. Outcomes were also compared between diffuse cutaneous SSc (dcSSc) vs limited cutaneous SSc (lcSSc). Multivariable regression analysis was performed to identify factors influencing GPH and GMH scores in SSc.ResultsA total of 10,502 complete responses from 276 SSc, 6006 non-SSc AIRDs, 545 NRAIDs, and 3675 controls as of May 2022 were included in the analysis. Respondents with SSc were older [SSc vs. non-SSc AIRDs vs. NRAIDs vs. controls: 55 (14) vs. 51 (15) vs. 45 (14) vs. 40 (14) years old, mean (SD), p < 0.001]. Among patients with SSc, 129 (47%) had dcSSc and 147 (53%) had lcSSc. SSc patients reported a significantly higher prevalence of ILD [SSc vs. non-SSc AIRDs vs. NRAIDs vs. controls: 30.4% vs. 5.5% vs. 1.5% vs. 0.2%, p < 0.001], and treatment with MMF [SSc vs. non-SSc AIRDs vs. NRAIDs vs. controls: 26.4% vs. 9.5% vs. 1.1% vs. 0%, p < 0.001].Patients with SSc had lower GPH and PROMIS PF-10a scores [SSc vs. non-SSc AIRDs vs. NRAIDs vs. controls: 13 (11–15) vs. 13 (11–15) vs. 15 (13–17) vs. 17 (15–18), median (IQR), p < 0.001;39 (33–46) vs. 39 (32–45) vs. 47 (40–50) vs. 49 (45–50), p < 0.001, respectively] and higher Pain VAS and PROMIS Fatigue-4a scores compared to those with NRAIDs or controls [SSc vs. non-SSc AIRDs vs. NRAIDs vs. controls: 3 (2–5) vs. 3 (1–6) vs. 2 (0–4) vs. 0 (0–2), p < 0.001;11 (8–14) vs. 11 (8–14) vs. 9 (7–13) vs. 7 (4–10), p < 0.001, respectively]. Patients with AIDs including SSc had lower GMH scores compared to controls [SSc vs. non-SSc AIRDs vs. NRAIDs vs. controls: 12.5 (10–15) vs. 13 (10–15) vs. 13 (11–16) vs. 15 (13–17), p < 0.001].Among SSc patients, GPH, GMH, and PROMIS PF-10a scores were lower in dcSSc compared to lcSSc [dcSSc vs. lcSSc: 12 (10–14) vs. 14 (11–15), p < 0.001;12 (10-14) vs. 13 (10-15), p<0.001;38 (30–43) vs. 41 (34–47), p < 0.001, respectively]. Pain VAS and PROMIS Fatigue-4a scores were higher in dcSSc compared to lcSSc [4 (2–6) vs. 3 (1–5), p < 0.001;12 (8–15) vs. 9 (8–13), p < 0.001, respectively].The independent factors for lower GPH scores in SSc were older age, Asian ethnicity, glucocorticoid use, and higher pain and fatigue scales, while mental health disorders and higher pain and fatigue scales were independently associated with lower GMH scores.ConclusionIn a global cohort, patient-reported physical and mental health were significantly worse in patients with SSc in comparison to those with non-SSc AIDs and without AIDs. Our findings support the critical need for more attention to patient's subjective experiences including pain and fatigue to improve the HRQOL in patients with SSc.Reference[1]Fazal ZZ, Sen P, Joshi M, et al. COVAD survey 2 long-term outcomes: unmet need and protocol. Rheumatol Int. 2022;42: 2151–58.Acknowledgements:NIL.Disclosure of InterestsKeina Yomono: None declared, Yuan Li: None dec ared, Vahed Maroufy: None declared, Naveen Ravichandran: None declared, Akira Yoshida: None declared, Kshitij Jagtap: None declared, Tsvetelina Velikova Speakers bureau: Pfizer and AstraZeneca, Parikshit Sen: None declared, Lorenzo Cavagna: None declared, Vishwesh Agarwal: None declared, Johannes Knitza: None declared, Ashima Makol: None declared, Dey Dzifa: None declared, Carlos Enrique Toro Gutierrez: None declared, Tulika Chatterjee: None declared, Aarat Patel: None declared, Rohit Aggarwal Consultant of: Bristol Myers-Squibb, Pfizer, Genentech, Octapharma, CSL Behring, Mallinckrodt, AstraZeneca, Corbus, Kezar, Abbvie, Janssen, Kyverna Alexion, Argenx, Q32, EMD-Serono, Boehringer Ingelheim, Roivant, Merck, Galapagos, Actigraph, Scipher, Horizon Therepeutics, Teva, Beigene, ANI Pharmaceuticals, Biogen, Nuvig, Capella Bioscience, and CabalettaBio, Grant/research support from: Bristol Myers-Squibb, Pfizer, Genentech, Octapharma, CSL Behring, Mallinckrodt, AstraZeneca, Corbus, Kezar, Abbvie, Janssen, Kyverna Alexion, Argenx, Q32, EMD-Serono, Boehringer Ingelheim, Roivant, Merck, Galapagos, Actigraph, Scipher, Horizon Therepeutics, Teva, Beigene, ANI Pharmaceuticals, Biogen, Nuvig, Capella Bioscience, and CabalettaBio, Latika Gupta: None declared, Masataka Kuwana Speakers bureau: Abbvie, Asahi-Kasei, Astellas, Boehringer-Ingelheim, Chugai, Eisai, MBL, Mochida, Nippon Shinyaku, Ono Pharmaceuticals, Tanabe-Mitsubishi, Consultant of: Astra Zeneka, Boehringer-Ingelheim, Chugai, Corbus, GSK, Horizon, Tanabe-Mitsubishi, Grant/research support from: Boehringer-Ingelheim, Vikas Agarwal: None declared.

Editorial

[...]they review all U.S. public company tax inversions in Ireland from 2010 to 2014 and find that inversions do not create additional value from operations. Program in Business at Chung Yuan Christian University, and University Chair Professor and department chair of the Department of Management Information Systems at the National Chengchi University (NCCU), Taiwan. Based on the research areas, he authored/co-authored over 80 refereed journal articles (e.g., Information & Management, Decision Support Systems, Journal of Information Systems, Information Systems Management, Communication of AIS, Journal of Global Information Management, Information System Frontiers) and published ten textbooks.

A SARS-CoV-2 neutralizing antibody protects from lung pathology in a COVID-19 hamster model (preprint)

The emergence of SARS-CoV-2 led to pandemic spread of coronavirus disease 2019 (COVID-19), manifesting with respiratory symptoms and multi-organ dysfunction. Detailed characterization of virus-neutralizing antibodies and target epitopes is needed to understand COVID-19 pathophysiology and guide immunization strategies. Among 598 human monoclonal antibodies (mAbs) from ten COVID-19 patients, we identified 40 strongly neutralizing mAbs. The most potent mAb CV07-209 neutralized authentic SARS-CoV-2 with IC50 of 3.1 ng/ml. Crystal structures of two mAbs in complex with the SARS-CoV-2 receptor-binding domain at 2.55 and 2.70 A revealed a direct block of ACE2 attachment. Interestingly, some of the near-germline SARS-CoV-2 neutralizing mAbs reacted with mammalian self-antigens. Prophylactic and therapeutic application of CV07-209 protected hamsters from SARS-CoV-2 infection, weight loss and lung pathology. Our results show that non-self-reactive virus-neutralizing mAbs elicited during SARS-CoV-2 infection are a promising therapeutic strategy.

Impact of policy interventions and social distancing on SARS-CoV-2 transmission in the United States (preprint)

Background: Policymakers have employed various non-pharmaceutical interventions (NPIs) such as stay-at-home orders and school closures to limit the spread of Coronavirus disease (COVID-19). However, these measures are not without cost, and careful analysis is critical to quantify their impact on disease spread and guide future initiatives. This study aims to measure the impact of NPIs on the effective reproductive number (Rt) and other COVID-19 outcomes in U.S. states. Methods: In order to standardize the stage of disease spread in each state, this study analyzes the weeks immediately after each state reached 500 cases. The primary outcomes were average Rt in the week following 500 cases and doubling time from 500 to 1000 cases. Linear and logistic regressions were performed in R to assess the impact of various NPIs while controlling for population density, GDP, and certain health metrics. This analysis was repeated for deaths with doubling time from 50 to 100 deaths and included several healthcare infrastructure control variables. Results: States that had a stay-at-home order in place at the time of their 500th case are associated with lower average Rt the following week compared to states without a stay-at-home order (p < 0.001) and are significantly less likely to have an Rt>1 (OR 0.07, 95% CI 0.01 to 0.37, p = 0.004). These states also experienced a significantly longer doubling time from 500 to 1000 cases (HR 0.35, 95% CI 0.17 to 0.72, p = 0.004). States in the highest quartile of average time spent at home were also slower to reach 1000 cases than those in the lowest quartile (HR 0.18, 95% CI 0.06 to 0.53, p = 0.002). Discussion: Few studies have analyzed the effect of statewide stay-at-home orders, school closures, and other social distancing measures in the U.S., which has faced the largest COVID-19 case burden. States with stay-at-home orders have a 93% decrease in the odds of having a positive Rt at a standardized point in disease burden. States that plan to scale back such measures should carefully monitor transmission metrics. Key words: COVID-19, SARS-CoV-2, Coronavirus, Public Policy, Social Distancing, Non-pharmaceutical Interventions, Stay-at-home Order, Shelter-in-place.

Validating a Widely Implemented Deterioration Index Model Among Hospitalized COVID-19 Patients (preprint)

IntroductionThe Epic Deterioration Index (EDI) is a proprietary prediction model implemented in over 100 U.S. hospitals that was widely used to support medical decision-making during the COVID-19 pandemic. The EDI has not been independently evaluated, and other proprietary models have been shown to be biased against vulnerable populations. MethodsWe studied adult patients admitted with COVID-19 to non-ICU care at a large academic medical center from March 9 through May 20, 2020. We used the EDI, calculated at 15-minute intervals, to predict a composite outcome of ICU-level care, mechanical ventilation, or in-hospital death. In a subset of patients hospitalized for at least 48 hours, we also evaluated the ability of the EDI to identify patients at low risk of experiencing this composite outcome during their remaining hospitalization. ResultsAmong 392 COVID-19 hospitalizations meeting inclusion criteria, 103 (26%) met the composite outcome. Median age of the cohort was 64 (IQR 53-75) with 168 (43%) African Americans and 169 (43%) women. Area under the receiver-operating-characteristic curve (AUC) of the EDI was 0.79 (95% CI 0.74-0.84). EDI predictions did not differ by race or sex. When exploring clinically-relevant thresholds of the EDI, we found patients who met or exceeded an EDI of 68.8 made up 14% of the study cohort and had a 74% probability of experiencing the composite outcome during their hospitalization with a median lead time of 24 hours from when this threshold was first exceeded. Among the 286 patients hospitalized for at least 48 hours who had not experienced the composite outcome, 14 (13%) never exceeded an EDI of 37.9, with a negative predictive value of 90% and a sensitivity above this threshold of 91%. ConclusionWe found the EDI identifies small subsets of high- and low-risk COVID-19 patients with fair discrimination. We did not find evidence of bias by race or sex. These findings highlight the importance of independent evaluation of proprietary models before widespread operational use among COVID-19 patients.

Multivariate Analysis of Factors Affecting COVID-19 Case and Death Rate in U.S. Counties: The Significant Effects of Black Race and Temperature (preprint)

Objectives: Coronavirus disease-19 (COVID-19) has spread rapidly around the world, and many risk factors including patient demographics, social determinants of health, environmental variables, underlying health conditions, and adherence to social distancing have been hypothesized to affect case and death rates. However, little has been done to account for the potential confounding effects of these factors. Using a large multivariate analysis, this study illuminates modulators of COVID-19 incidence and mortality in U.S. counties while controlling for risk factors across multiple domains. Methods: Data on COVID-19 and various risk factors in all U.S. counties was collected from publicly available data sources through April 14, 2020. Counties with at least 50 COVID-19 cases were included in case analyses and those with at least 10 deaths were included in mortality models. The 661 counties meeting inclusion criteria for number of cases were grouped into quartiles and comparisons of risk factors were made using t-tests between the highest and lowest quartiles. Similar comparisons for 217 counties were made for above average and below average deaths/100,000. Adjusted linear and logistic regression analyses were performed to evaluate the independent effects of factors that significantly impacted cases and deaths. Results: Univariate analyses demonstrated numerous significant differences between cohorts for both cases and deaths. Risk factors associated with increased cases and/or deaths per 100,000 included increased GDP per capita, decreased social distancing, increased age, increased percent Black, decreased percent Hispanic, decreased percent Asian, decreased health, increased poverty, increased diabetes, increased coronary heart disease, increased physical inactivity, increased alcohol consumption, increased tobacco use, and decreased access to primary care. Multivariate regression analyses demonstrated Black race is a risk factor for worse COVID-19 outcome independent of comorbidities, poverty, access to health care, and other mitigating factors. Lower daily temperatures was also an independent risk factor in case load but not deaths. Conclusions: U.S. counties with a higher proportion of Black residents are associated with increased COVID-19 cases and deaths. However, the various suggested mechanisms, such as socioeconomic and healthcare predispositions, did not appear to drive the effect of race in our model. Counties with higher average daily temperatures are also associated with decreased COVID-19 cases but not deaths. Several theories are posited to explain these findings, including prevalence of vitamin D deficiency. Additional studies are needed to further understand these effects.

Analysis of Potential Risk of COVID-19 Infections in China Based on a Pairwise Epidemic Model (preprint)

The ongoing outbreak of the novel coronavirus pneumonia (also known as COVID-19) has triggered a series of stringent control measures in China, such as city closure, traffic restrictions, contact tracing and household quarantine. These containment efforts often lead to changes in the contact pattern among individuals of the population. Many existing compartmental epidemic models fail to account for the effects of contact structure. In this paper, we devised a pairwise epidemic model to analyze the COVID-19 outbreak in China based on confirmed cases reported during the period February 3rd--17th, 2020. By explicitly incorporating the effects of family clusters and contact tracing followed by household quarantine and isolation, our model provides a good fit to the trajectory of COVID-19 infections and is useful to predict the epidemic trend. We obtained the average of the reproduction number $R=1.494$ ($95\%$ CI: $1.483-1.507$) for Hubei province and $R=1.178$ ($95\%$ CI: $1.145-1.158$) for China (except Hubei), suggesting that some existing studies may have overestimated the reproduction number by neglecting the dynamical correlations and clustering effects. We forecasted that the COVID-19 epidemic would peak on February 13th ($95\%$ CI: February $9-17$th) in Hubei and 6 days eariler in the regions outside Hubei. Moreover the epidemic was expected to last until the middle of March in China (except Hubei) and late April in Hubei. The sensitivity analysis shows that ongoing exposure for the susceptible and population clustering play an important role in the disease propagation. With the enforcement of household quarantine measures, the reproduction number $R$ effectively reduces and epidemic quantities decrease accordingly. Furthermore, we gave an answer to the public concern on how long the stringent containment strategies should maintain. Through numerical analysis, we suggested that the time for the resumption of work and production in China (except Hubei) and Hubei would be the middle of March and the end of April, 2020, respectively. These constructive suggestions may bring some immeasurable social-economic benefits in the long run.

Radiological Findings in COVID-19 and Adaptive Approaches for Radiology Departments: Literature Review and Experience Sharing

Radiological investigations play an important role in the treatment course of patients with coronavirus disease 2019 (COVID-19) and radiologists should be familiar with the imaging characteristics. Being an integral component of the healthcare system, radiology departments have made adaptations to enhance infection control and strengthen the service. In this article, we review the radiological features of COVID-19 on chest radiography and computed tomography, and share experiences on the adaptive approach of radiology departments amidst the COVID-19 pandemic.

The prevalence and risk factors for anxiety in frontline nurses under COVID-19 pandemic based on a large cross-sectional study using the propensity score-matched method

Introduction: We determined the prevalence of anxiety and the associated risk factors in frontline nurses under COVID-19 pandemic. Methods: This cross-sectional study was conducted from February 20, 2020, to March 20, 2020, and involved 562 frontline nurses. The effective response rate was 87.68%. After propensity score matched, there were 532 participants left. Extensive characteristics, including demographics, dietary habits, life-related factors, work-related factors, and psychological factors were collected based on a self-reported questionnaire. Specific scales measured the levels of sleep quality, physical activity, anxiety, perceived organization support and psychological capital. Adjusted odds ratios and 95% confidence intervals were determined by binary paired logistic regression. Results: Of the nurses enrolled in the study, 33.60% had anxiety. Five independent risk factors were identified for anxiety: poor sleep quality (OR=1.235), experienced major events (OR=1.653), lower resilience and optimism of psychological capital (OR=0.906, and OR=0.909) and no visiting friend constantly (OR=0.629). Conclusions: This study revealed a considerable high prevalence of anxiety in frontline nurses during the COVID-19 outbreak, and identified five risk factors, which were poor sleep quality, experienced major events, lower resilience and optimism of psychological capital, and no visiting friend constantly. Protecting mental health of nurses is important for COVID-19 pandemic control and their wellbeing. These findings enrich the existing theoretical model of anxiety and demonstrated a critical need for additional strategies that could address the mental health in frontline nurses for policymakers.