A Predictive Model of COVID-19 mRNA Vaccine Reactivity based on Dual Attention for Federated Learning Scenarios

The base reactivity of the mRNA sequence has a significant impact on the effectiveness of the mRNA vaccine in fighting against the pandemic of COVID-19. The annotation of mRNA sequence reactivity value is a time-consuming and labor-intensive work, which belongs to the private digital assets of each medical institution. It is not practical to train a predictive model by pooling private data from various parties. Fortunately, federated learning techniques can serve to collaboratively train a predictive model among medical institutions while preserving respective digital assets. However, due to the scarcity of data from each participant, conventional sequential prediction mod-els often fail to perform well. To overcome such a challenge, we propose a reactivity value prediction model based on both the self-attention and the convolutional attention mechanisms only requiring a small dataset of labeled samples. Inspired by BERT, we first train a self-attention feature extraction model through self-supervision using both labeled and unlabeled mRNA samples. In this way, the information of mRNA in the semantic space is deeply mined. Then, a convolutional attention block follows the self-attention block, to extract the attention matrix from the base-pair probability matrix and adjacency matrix. By doing so, the attention matrix can compensate for the insensitivity of the self-attention mechanism to the spatial information of mRNA. By using the Open Vaccine RNA database, experiments show that our prediction model for unseen mRNA has a better performance than other state-of-the-art deep learning models that are used to process gene sequences. Further ablation experiments demonstrate that the existence of the dual attention mechanism reduces the risk of overfitting, resulting in an excellent generalization capability of our model. © 2022 IEEE.

Comprehensive, longitudinal wastewater surveillance for SARS-CoV-2 across a university campus, demonstrates low levels of SARS-CoV-2 activity relative to the surrounding community

Background. Universities are interactive communities where frequent contacts between individuals occur, increasing the risk of outbreaks of COVID-19. We embarked upon a real-time wastewater (WW) monitoring program across the University of Calgary (UofC) campus measuring WW SARS-CoV-2 burden relative to levels of disease in the broader surrounding community. Figure 1 The colour scheme shows 6 sewer sub-catchments at the University of Calgary. Auto samplers were deployed at 4 sampling nodes within sub-catchments CR and YA (both residence halls), and UCE and UCS (catchments that include several campus buildings). Figure 2 Log10-transformed abundance (i.e., copies per mL) of nucleocapsid gene (i.e., N1) for SARS-CoV-2 for each sampling location during October 2021 - April 2022. Locations denoted by the same letters (A, B, or C) show no statistical difference (p > 0.05) according to the Wilcoxon rank-sum test. The WWTP sample corresponds to a catchment area covering most of Calgary including the university campus, for which sampling locations CR, UCE, UCS, and UCW are defined in Fig. 1. Methods. From October 2021 - April 2022, WW was collected thrice weekly across UofC campus through 4 individual sewer sampling nodes (Fig. 1) using autosamplers (C.E.C. Analytics, CA). Results from these 4 nodes were compared with community monitoring at Calgary's largest WW treatment plant (WWTP), which received WW from surrounding neighborhoods, and also from UofC. Nucleic acid was extracted from WW for RTqPCR quantification of the N1 nucleocapside gene from SARS-CoV-2 genomic RNA. Qualitative (positive samples defined if cycle threshold < 40) and quantitative statistical analyses were performed using R. Results. Levels of SARS-CoV-2 in WW were significantly lower at all campus monitoring sites relative to the WWTP (Wilcoxon rank-sum test p < 0.05;Fig. 2). The proportion of WW samples that were positive for SARS-CoV-2 was significantly higher for WWTP than at least two campus locations (p < 0.05 for Crowsnest Hall and UCE - University way and campus drive) according to Fischer's exact 2-sided test. The proportion of WW samples with positive WW signals were still higher for WWTP than the other two locations, but statistically not significant (p = 0.216). Among campus locations, the buildings in UCE catchment showed much lower N1 signals than other catchments, likely owing to buildings in this catchment primarily being administration and classroom environments, with lower human-to-human contact and less defecation compared to the other 3 catchments, which include residence hall, a dining area, and/or laboratory spaces. Conclusion. Our results show that SARS-CoV-2 RNA shedding in WW at the U of C is significantly lower than the city-wide signal associated with surrounding neighborhoods. Furthermore, we demonstrate that WW testing at well-defined nodes is a sampling strategy for potentially locating specific places where high transmission of infectious disease occurs.

Surveillance of SARS-CoV-2 variants of concern (VOC) in hospital wastewater (WW) and its correlation with hospitalized cases of COVID-19 and the occurrence of outbreaks

Background. WW surveillance enables real time monitoring of SARS-CoV-2 burden in defined sewer catchment areas. Here, we assessed the occurrence of total, Delta and Omicron SARS-CoV-2 RNA in sewage from three tertiary-care hospitals in Calgary, Canada. Methods. Nucleic acid was extracted from hospital (H) WW using the 4S-silica column method. H-1 and H-2 were assessed via a single autosampler whereas H-3 required three separate monitoring devices (a-c). SARS-CoV-2 RNA was quantified using two RT-qPCR approaches targeting the nucleocapsid gene;N1 and N200 assays, and the R203K/G204R and R203M mutations. Assays were positive if Cq< 40. Cross-correlation function analyses (CCF) was performed to determine the timelagged relationships betweenWWsignal and clinical cases. SARS-CoV-2 RNA abundance was compared to total hospitalized cases, nosocomial-acquired cases, and outbreaks. Statistical analyses were conducted using R. Results. Ninety-six percent (188/196) of WW samples collected between Aug/ 21-Jan/22 were positive for SARS-CoV-2. Omicron rapidly supplanted Delta by mid-December and this correlated with lack of Delta-associated H-transmissions during a period of frequent outbreaks. The CCF analysis showed a positive autocorrelation between the RNA concentration and total cases, where the most dominant cross correlations occurred between -3 and 0 lags (weeks) (Cross-correlation values: 0.75, 0.579, 0.608, 0.528 and 0.746 for H-1, H-2, H-3a, H-3b and H-3c;respectively). VOC-specific assessments showed this positive association only to hold true for Omicron across all hospitals (cross-correlation occurred at lags -2 and 0, CFF value range between 0.648 -0.984). We observed a significant difference in median copies/ ml SARS-CoV-2 N-1 between outbreak-free periods vs outbreaks for H-1 (46 [IQR: 11-150] vs 742 [IQR: 162-1176], P< 0.0001), H-2 (24 [IQR: 6-167] vs 214 [IQR: 57-560], P=0.009) and H-3c (2.32 [IQR: 0-19] vs 129 [IQR: 14-274], P=0.001). Conclusion. WWsurveillance is a powerful tool for early detection andmonitoring of circulating SARS-CoV-2VOCs.Total SARS-CoV-2 andVOC-specificWWsignal correlated with hospitalized prevalent cases of COVID-19 and outbreak occurrence.

A CASE OF ANTISYNTHETASE SYNDROME: THE IMPORTANCE OF HISTORY IN DISTINGUISHING ETIOLOGY OF INTERSTITIAL LUNG DISEASE IN A COVID-19 WORLD

SESSION TITLE: Systemic Diseases Causing Pulmonary Havoc SESSION TYPE: Rapid Fire Case Reports PRESENTED ON: 10/18/2022 10:15 am - 11:10 am INTRODUCTION: In the coronavirus disease 2019 (COVID-19) era, the etiology of interstitial lung disease (ILD) should remain broad to ensure accurate diagnosis and the proper treatment of patients. Vital to the art of medicine is taking a comprehensive history, and anchoring on a common diagnosis such as COVID-19 can result in early dismissal of alternate etiologies that physicians have an obligation to explore. CASE PRESENTATION: A 58-year-old male with a history of diabetes, hypothyroidism, and hypertension presented to the emergency department (ED) with dyspnea and fever. Initial CT chest imaging was significant for reticular and fibrotic changes with peripheral ground-glass and solid nodular opacities, some with areas of central clearing. Despite negative PCR testing, he was diagnosed with COVID-19 and discharged on oxygen with pulmonary follow-up. He continued to have arthralgias, proximal muscle weakness, low-grade fevers, and weight loss. He re-presented to the ED and was admitted for hypovolemia and further exploration into a potential autoimmune etiology of his symptoms. Labs were significant for a creatine kinase of 3,381 U/L, positive autoimmune antibodies [ANA (1:320), Jo-1 (>8.0 U), and SS-A/Ro (1.4 U)], and elevated ESR and CRP (30 mm/hr and 81 mg/L). Repeat CT revealed persistent parenchymal changes. Bronchoscopy was performed without anatomical abnormalities, and bronchoalveolar lavage (BAL) fluid was normal in appearance and negative for infectious etiologies. Though the patient was a farmer and possessed risk factors for hypersensitivity pneumonitis, lack of lymphocytic predominance on BAL, negative hypersensitivity panel, and uncharacteristic CT findings helped exclude this diagnosis. The patient was diagnosed with antisynthetase syndrome and treated with pulse dose intravenous solumedrol before transitioning to prednisone with resolution of muscle weakness and radiographic improvement in lung infiltrates. Muscle biopsy was deferred given the rapid clinical response and serum markers consistent with the diagnosis. DISCUSSION: Antisynthetase syndrome is a rare cause of ILD and often presents with myositis, arthritis, skin changes, Raynaud's phenomenon, and fever [1]. These symptoms, combined with the aminoacyl-tRNA synthetase antibody—most commonly the Jo-1 antibody—help confirm the diagnosis [2]. Due to a lack of established diagnostic criteria, muscle biopsy is often used to exclude other causes of myositis [3]. The ILD associated with antisynthetase syndrome is a significant cause of morbidity and mortality, and delay in diagnosis can lead to progression of lung injury. CONCLUSIONS: Chest imaging findings in COVID-19 are nonspecific, and post-COVID lung disease often presents similarly to other ILDs [1]. Because of this, history and physical exam remain crucial tools to reflect on alternate diagnoses for ILD and will continue to be necessary as we evolve through this COVID-19 era. Reference #1: Devi HG, Pasha MM, Padmaja MS, Halappa S. Antisynthetase Syndrome: A Rare Cause for ILD. Journal of Clinical and Diagnostic Research : JCDR. 2016;10(3):OD08. doi:10.7860/JCDR/2016/16872.7361 Reference #2: Cavagna L, Trallero-Araguás E, Meloni F, et al. Influence of Antisynthetase Antibodies Specificities on Antisynthetase Syndrome Clinical Spectrum Time Course. Journal of Clinical Medicine. 2019;8(11). doi:10.3390/jcm8112013 Reference #3: Schmidt J. Current Classification and Management of Inflammatory Myopathies. J Neuromuscul Dis. 2018;5(2):109-129. doi: 10.3233/JND-180308. PMID: 29865091;PMCID: PMC6004913. DISCLOSURES: No relevant relationships by Dustin Norton No relevant relationships by Alyssa Simon No relevant relationships by Kang Rui Xiang

A Transfer Learning Image Classification Method Using Self-Supervised Information

Research on intelligent diagnosis and treatment is a major frontier issue in the current era of medical big data. For the global health crisis COVID-19, the radiological imaging techniques CT can provide useful and important information thus widely preferred due to its merit and three-dimensional view of the lung. However, to classify the CT-slices to assist in diagnosis, due to the annotation by radiologists is a highly subjective task, tedious and time-consuming work often influenced by individual bias and clinical experiences. Moreover, the current image classification methods cannot work well on the massive real-Time totally unlabeled CT scans. To address these challenges, we proposed a transfer learning method using self-supervised information to classify the unlabeled CT images, using an auxiliary task of segmentation to improve classification efficiency. We classified the totally unlabeled CT scans from Huoshenshan Hospital into ordinary, severe and critical cases, and the accuracy rate reached 86%. The experimental results show that the use of small-sample semi-supervised transfer learning algorithm can be used in insufficient CT images. Our framework can improve the learning ability and achieve a higher performance. Extensive experiments on real CT volumes demonstrate that the proposed method outperforms most current models and advances the state-of-The-Art performance. © 2022 IEEE.

Trans-Inf-Net: COVID-19 Lung Infection Segmentation Based on Transformer

For the global health crisis COVID-19, the radio-logical imaging techniques CT have demonstrated effectiveness in both current diagnosis and evaluation of disease evolution. However, the manual delineation of lung infections is tedious and time-consuming work, and infection annotation by radiologists is a highly subjective task, often influenced by individual bias and clinical experiences. To address these challenges, we proposed a transformer learning method (Trans-Inf-Net) to automatically identify infected regions from chest CT slices. In our Trans-Inf-Net, a parallel partial decoder is used to aggregate the high-level features and generate a global map. Then, the implicit reverse attention and explicit edge-Attention are utilized to model the boundaries and enhance the representations. Moreover, to alleviate the shortage of labeled data, we present a segmentation framework based on a randomly selected propagation strategy and transformer, which only requires a few labeled images and leverages primarily unlabeled data. We apply attention in conjunction with convolutional networks, while keeping their overall structure in place. a pure transformer applied directly to sequences of image patches can perform very well on image segmentation tasks. Our framework can improve the learning ability and achieve a higher performance. Extensive experiments on COVID-SemiSeg and real CT volumes demonstrate that the proposed Trans-Inf-Net outperforms most cutting-edge segmen-Tation models and advances the state-of-The-Art performance. © 2022 IEEE.

Implementing, launching and sustaining a quality improvement initiative in the era of covid-19: Challenges and successes

Introduction: Only 60% of atrial fibrillation (AF) patients with elevated stroke risk receive anticoagulation (AC). Closing this gap in AC use is particularly challenging in the face of the COVID19 pandemic. Electronic health record (EHR) alerts integrated into in-person and telehealth visits have the potential to close the gap. Hypothesis: A triad approach consisting of interventions at the practice provider at patient level could improve anticoagulation rates in patients with atrial fibrillation. Methods: In collaboration with the Heart Rhythm Society Quality Improvement Committee and UMass, University of Florida (UFL) Jacksonville launched a 2020 quality improvement (QI) intervention, including several Plan-Do-Study-Act (PDSA) cycles, centered around an outpatient, electronic health record (EHR) alert linked to an order set for AC meds, labs, and specialty referrals. The alert fired when cardiologists or primary care physicians saw AF patients in clinic with a CHA2DS2-VASc score of ≥2 who were not on AC. Due to COVID-19, several of the PDSA interventions required adjustments due to redeployment of information technology staff mobilized to generate COVID-19 reports, a pivot for clinicians and patients to telehealth visits, and a change in clinician and patient priorities from routine cardiovascular/preventive care to COVID-19 diagnosis and prevention. To assess intervention effectiveness, the change in AC use as a function of time was measured using a weighted least squares linear regression. Results: At time of launch, 2357 of 3555 eligible patients (56.3%) were on AC. At study end, six months later, the percentage of patients on AC increased by 1.5% to 57.8 %. Based on the population of untreated patients at UFL and the stroke rate for untreated patients (available from large registry data), and assuming absence of any competing/secular trend to explain the growth in AC use, we calculated that a 1.5% increase in the AC treatment percentage could result in the prevention of 1.5 strokes over one year. Conclusion: We demonstrated that an EHR alert can raise the rate of AC use in patients with AF after several rounds of PDSAs. In future efforts, we plan to reassess the AC percentage in our population and confirm the sustainability of our QI efforts as attention focuses back from the pandemic to routine cardiovascular/ preventive care.

Information anxiety, intergroup emotion, and rational coping in hotel employees under normalized pandemic prevention measures

This study applied the intergroup emotion theory framework to explore the internal driving framework of hotel employees' information anxiety and the internal correlation mechanism of intergroup emotions and coping actions under normal epidemic prevention conditions. In this study, a mixed research method was used. Based on 105 videos and 15 in-depth interview records, the internal driving framework of hotel employees' information anxiety identified in the qualitative research context and the internal correlation mechanism of hotel employees’ information anxiety, analysed within the framework of intergroup sentiment analysis, was examined based on 213 valid questionnaires. The results verified the internal relationship between information anxiety of hotel employees and behaviour tendency and intergroup relationship, and also confirmed information anxiety of employees as a mediating variable on intergroup relationship and cognitive evaluation. © 2021 The Authors