Towards Algorithmic Fidelity: Mental Health Representation across Demographics in Synthetic vs. Human-generated Data (preprint)

Synthetic data generation has the potential to impact applications and domains with scarce data. However, before such data is used for sensitive tasks such as mental health, we need an understanding of how different demographics are represented in it. In our paper, we analyze the potential of producing synthetic data using GPT-3 by exploring the various stressors it attributes to different race and gender combinations, to provide insight for future researchers looking into using LLMs for data generation. Using GPT-3, we develop HEADROOM, a synthetic dataset of 3,120 posts about depression-triggering stressors, by controlling for race, gender, and time frame (before and after COVID-19). Using this dataset, we conduct semantic and lexical analyses to (1) identify the predominant stressors for each demographic group; and (2) compare our synthetic data to a human-generated dataset. We present the procedures to generate queries to develop depression data using GPT-3, and conduct analyzes to uncover the types of stressors it assigns to demographic groups, which could be used to test the limitations of LLMs for synthetic data generation for depression data. Our findings show that synthetic data mimics some of the human-generated data distribution for the predominant depression stressors across diverse demographics.

Change of Heart: Virtual Embryonic 3D Heart Model with Interchangeable Conotruncal Septa Promotes Clinically Oriented Learning

Teaching and learning embryology pose challenges due to rapid and complex developmental processes occurring in four-dimensions especially without effective visual aids. Previous research studies have shown that virtual and 3D printed embryo models increase learning outcomes and interest in embryology. One of the most challenging and clinically relevant concepts in embryology is the heart outflow tract septation involving 180-degree spatial rotation and precise alignment of the growing septa that can often go awry, resulting in various congenital cardiac septal defects. Unfortunately, there are no visual aids that effectively demonstrate this complex process in 3D. To that end, a 3D embryonic virtual heart development model (EVH) with interchangeable conotruncal septa (Figure 1) was created and its educational efficacy was tested in a COMIRB exempt (# 20-2124), randomized single blind study with the first-year medical students completing a flipped-classroom embryology active learning event occurring in a virtual platform due to COVID-19 restrictions. As a part of the course requirement, all students first watched a pre-recorded lecture video and completed a quiz on their own before attending the synchronous virtual active learning event. During the active learning event, students were randomized into control or experimental breakout session groups. Experimental breakout session groups (16 groups, n=80) were provided access to the EVH with accompanying prompts designed to reinforce 3D spatial orientation and congenital septal defects demonstrated by the EVH (Figure 2). The control groups (14 groups, n=70) were tasked to complete a similar learning activity but on a different developmental process. All students completed a second quiz and a survey after the active learning event. A Kruskal-Wallis test comparing normalized pre-quiz versus post-quiz results revealed a significant increase on the post-quiz performance in the experimental group (p = 0.0198, df = 3) but not in the control group (p>0.05). Descriptive statistics on quantitative survey data revealed that students positively perceived the educational value of the EVH, especially for demonstrating 3D changes of the normally and abnormally developing outflow tract. Consistently, thematic analysis of the survey comments indicated that the spatial and visual aspects of the EVH demonstrating the congenital defects were perceived as essential. The significantly high learning outcomes immediately after the resource interaction with EVH and the favorability of the EVH warrant a further development and research. Group and virtual classroom effects are other confounds that should be reduced in future studies.

Derivation and validation of a triage tool for acutely ill adults with suspected COVID-19: The PRIEST observational cohort study (preprint)

ObjectivesWe aimed to derive and validate a triage tool, based on clinical assessment alone, for predicting adverse outcome in acutely ill adults with suspected COVID-19 infection. MethodsWe undertook a mixed prospective and retrospective observational cohort study in 70 emergency departments across the United Kingdom (UK). We collected presenting data from 22445 people attending with suspected COVID-19 between 26 March 2020 and 28 May 2020. The primary outcome was death or organ support (respiratory, cardiovascular, or renal) by record review at 30 days. We split the cohort into derivation and validation sets, developed a clinical score based on the coefficients from multivariable analysis using the derivation set, and the estimated discriminant performance using the validation set. ResultsWe analysed 11773 derivation and 9118 validation cases. Multivariable analysis identified that age, sex, respiratory rate, systolic blood pressure, oxygen saturation/inspired oxygen ratio, performance status, consciousness, history of renal impairment, and respiratory distress were retained in analyses restricted to the ten or fewer predictors. We used findings from multivariable analysis and clinical judgement to develop a score based on the NEWS2 score, age, sex, and performance status. This had a c-statistic of 0.80 (95% confidence interval 0.79-0.81) in the validation cohort and predicted adverse outcome with sensitivity 0.98 (0.97-0.98) and specificity 0.34 (0.34-0.35) for scores above four points. ConclusionA clinical score based on NEWS2, age, sex, and performance status predicts adverse outcome with good discrimination in adults with suspected COVID-19 and can be used to support decision-making in emergency care. RegistrationISRCTN registry, ISRCTN28342533, http://www.isrctn.com/ISRCTN28342533

SARS-CoV-2 spike D614G variant exhibits highly efficient replication and transmission in hamsters (preprint)

SARS-CoV-2 causes disease varying in severity from asymptomatic infections to severe respiratory distress and death in humans. The viral factors which determine transmissibility and pathogenicity are not yet clearly characterized. We used the hamster infection model to compare the replication ability and pathogenicity of five SARS-CoV-2 strains isolated from early cases originating in Wuhan, China, in February, and infected individuals returning from Europe and elsewhere in March 2020. The HK-13 and HK-95 isolates showed distinct pathogenicity in hamsters, with higher virus titers and more severe pathological changes in the lungs observed compared to other isolates. HK-95 contains a D614G substitution in the spike protein and demonstrated higher viral gene expression and transmission efficiency in hamsters. Intra-host diversity analysis revealed that further quasi species were generated during hamster infections, indicating that strain-specific adaptive mutants with advantages in replication and transmission will continue to arise and dominate subsequent waves of SARS-CoV-2 dissemination.

Pathogenicity, immunogenicity, and protective ability of an attenuated SARS-CoV-2 variant with a deletion at the S1/S2 junction of the spike protein (preprint)

SARS-CoV-2 is zoonotic origin and contains a PRRA polybasic cleavage motif which is considered critical for efficient infection and transmission in humans. We previously reported on a panel of attenuated SARS-CoV-2 variants with deletion at the S1/S2 junction of spike protein. Here we characterize pathogenicity, immunogenicity, and protective ability of a further cell-adapted SARS-CoV-2 variant, Ca-DelMut, in in vitro and in vivo systems. Ca-DelMut replicates more efficiently than wild type or parental virus in cells, but causes no apparent disease in hamsters, despite replicating in respiratory tissues. Unlike wild type virus, Ca-DelMut causes no apparent pathological changes and does not induce elevated proinflammatory cytokines in hamster infections, but still triggers a strong neutralizing antibody response in hamsters. Ca-DelMut immunized hamsters challenged with wild type SARS-CoV-2 are fully protected with no sign of virus replication in the upper or lower respiratory tract of challenged animals, demonstrating sterilizing immunity.

Prognostic accuracy of emergency department triage tools for adults with suspected COVID-19: The PRIEST observational cohort study (preprint)

ObjectivesThe World Health Organisation (WHO) and National Institute for Health and Care Excellence (NICE) recommend various triage tools to assist decision-making for patients with suspected COVID-19. We aimed to estimate the accuracy of triage tools for predicting severe illness in adults presenting to the emergency department (ED) with suspected COVID-19 infection. MethodsWe undertook a mixed prospective and retrospective observational cohort study in 70 EDs across the United Kingdom (UK). We collected data from people attending with suspected COVID-19 between 26 March 2020 and 28 May 2020, and used presenting data to determine the results of assessment with the following triage tools: the WHO algorithm, NEWS2, CURB-65, CRB-65, PMEWS and the swine flu adult hospital pathway (SFAHP). We used 30-day outcome data (death or receipt of respiratory, cardiovascular or renal support) to determine prognostic accuracy for adverse outcome. ResultsWe analysed data from 20892 adults, of whom 4672 (22.4%) died or received organ support (primary outcome), with 2058 (9.9%) receiving organ support and 2614 (12.5%) dying without organ support (secondary outcomes). C-statistics for the primary outcome were: CURB-65 0.75; CRB-65 0.70; PMEWS 0.77; NEWS2 (score) 0.77; NEWS2 (rule) 0.69; SFAHP (6-point) 0.70; SFAHP (7-point) 0.68; WHO algorithm 0.61. All triage tools showed worse prediction for receipt of organ support and better prediction for death without organ support. At the recommended threshold, PMEWS and the WHO criteria showed good sensitivity (0.96 and 0.95 respectively), at the expense of specificity (0.31 and 0.27 respectively). NEWS2 showed similar sensitivity (0.96) and specificity (0.28) when a lower threshold than recommended was used. ConclusionCURB-65, PMEWS and NEWS2 provide good but not excellent prediction for adverse outcome in suspected COVID-19, and predicted death without organ support better than receipt of organ support. PMEWS, the WHO criteria and NEWS2 (using a lower threshold than usually recommended) provide good sensitivity at the expense of specificity. RegistrationISRCTN registry, ISRCTN28342533, http://www.isrctn.com/ISRCTN28342533

Prognostic accuracy of emergency department triage tools for children with suspected COVID-19: The PRIEST observational cohort study (preprint)

Objectives: Emergency department clinicians can use triage tools to predict adverse outcome and support management decisions for children presenting with suspected COVID-19. We aimed to estimate the accuracy of triage tools for predicting severe illness in children presenting to the emergency department (ED) with suspected COVID-19 infection. Methods: We undertook a mixed prospective and retrospective observational cohort study in 44 EDs across the United Kingdom (UK). We collected data from children attending with suspected COVID-19 between 26 March 2020 and 28 May 2020, and used presenting data to determine the results of assessment using the WHO algorithm, swine flu hospital pathway for children (SFHPC), Paediatric Observation Priority Score (POPS) and Childrens Observation and Severity Tool (COAST). We recorded 30-day outcome data (death or receipt of respiratory, cardiovascular or renal support) to determine prognostic accuracy for adverse outcome. Results: We collected data from 1530 children, including 26 (1.7%) with an adverse outcome. C-statistics were 0.80 (95% confidence interval 0.73-0.87) for the WHO algorithm, 0.80 (0.71-0.90) for POPS, 0.76 (0.67-0.85) for COAST, and 0.71 (0.59-0.82) for SFHPC. Using pre-specified thresholds, the WHO algorithm had the highest sensitivity (0.85) and lowest specificity (0.75), but POPS and COAST could optimise sensitivity (0.96 and 0.92 respectively) at the expense of specificity (0.25 and 0.38 respectively) by using a threshold of any score above zero instead of the pre-specified threshold. Conclusion: Existing triage tools have good but not excellent prediction for adverse outcome in children with suspected COVID-19. POPS and COAST could achieve an appropriate balance of sensitivity and specificity for supporting decisions to discharge home by considering any score above zero to be positive.

Characterisation of 22446 patients attending UK emergency departments with suspected COVID-19 infection: Observational cohort study (preprint)

BackgroundHospital emergency departments play a crucial role in the Initial management of suspected COVID-19 infection. We aimed to characterise patients attending emergency departments with suspected COVID-19, including subgroups based on sex, ethnicity and COVID-19 test results. MethodsWe undertook a mixed prospective and retrospective observational cohort study in 70 emergency departments across the United Kingdom (UK). We collected presenting data from 22446 people attending with suspected COVID-19 between 26 March 2020 and 28 May 2020. Outcomes were admission to hospital, COVID-19 result, organ support (respiratory, cardiovascular or renal), and death, by record review at 30 days. ResultsAdults were acutely unwell (median NEWS2 score 4) and had high rates of admission (67.1%), COVID-19 positivity (31.2%), organ support (9.8%) and death (15.9%). Children had much lower rates of admission (27.4%), COVID-19 positivity (1.2%), organ support (1.4%) and death (0.3%). Adult men and women presented in similar numbers (10210 versus 10506), but men were more likely to be admitted (72.9% v 61.4%), require organ support (12.2% v 7.7%) and die (18.7% v 13.3%). Black or Asian adults tended to be younger than White adults (median age 54, 50 and 67 years), were less likely to be admitted (60.8%, 57.3%, 69.6%) or die (11.9%, 11.2%, 16.8%), but were more likely to require organ support (15.9%, 14.3%, 8.9%) or have a positive COVID-19 test (40.8%, 42.1%, 30.0%). Adults admitted with confirmed COVID-19 had similar age and comorbidities (except chronic lung disease) to those who did not have COVID-19 confirmed, but were much more likely to need organ support (22.2% v 8.9%) or die (32.7% v 15.9%). ConclusionsImportant differences exist between patient groups presenting to the emergency department with suspected COVID-19. People with confirmed COVID-19 have a poor prognosis, compared with similar emergency admissions without confirmed COVID-19. RegistrationISRCTN registry, ISRCTN56149622, http://www.isrctn.com/ISRCTN28342533

Post-exertion oxygen saturation as a prognostic factor for adverse outcome in patients attending the emergency department with suspected COVID-19: Observational cohort study (preprint)

BackgroundMeasurement of post-exertion oxygen saturation has been proposed to assess illness severity in suspected COVID-19 infection. We aimed to determine the accuracy of post-exertional oxygen saturation for predicting adverse outcome in suspected COVID-19. MethodsWe undertook an observational cohort study across 70 emergency departments during first wave of the COVID-19 pandemic in the United Kingdom. We collected data prospectively, using a standardised assessment form, and retrospectively, using hospital records, from patients with suspected COVID-19, and reviewed hospital records at 30 days for adverse outcome (death or receiving organ support). Patients with post-exertion oxygen saturation recorded were selected for this analysis. ResultsWe analysed data from 817 patients with post-exertion oxygen saturation recorded after excluding 54 in whom measurement appeared unfeasible. The c-statistic for post-exertion change in oxygen saturation was 0.589 (95% confidence interval 0.465 to 0.713), and the positive and negative likelihood ratios of a 3% or more desaturation were respectively 1.78 (1.25 to 2.53) and 0.67 (0.46 to 0.98). Multivariable analysis showed that post-exertion oxygen saturation was not a significant predictor of adverse outcome when baseline clinical assessment was taken into account (p=0.368). Secondary analysis excluding patients in whom post-exertion measurement appeared inappropriate resulted in a c-statistic of 0.699 (0.581 to 0.817), likelihood ratios of 1.98 (1.26 to 3.10) and 0.61 (0.35 to 1.07), and some evidence of additional prognostic value on multivariable analysis (p=0.019). ConclusionsPost-exertion oxygen saturation provides modest prognostic information in the assessment of patients attending the emergency department with suspected COVID-19. RegistrationISRCTN registry, ISRCTN56149622, http://www.isrctn.com/ISRCTN28342533 Key messagesWhat is already known on this subject? O_LIPost exertional decrease in oxygen saturation can be used to predict prognosis in chronic lung diseases C_LIO_LIPost exertional desaturation has been proposed as a way of predicting adverse outcome in people with suspected COVID-19 C_LI What this study adds: O_LIPost-exertion oxygen saturation provides modest prognostic information in the assessment of patients attending the emergency department with suspected COVID-19 C_LI