Dataset Information

Development and prospective validation of COVID-19 chest X-ray screening model for patients attending emergency departments

ABSTRACT: Chest X-rays (CXRs) are the first-line investigation in patients presenting to emergency departments (EDs) with dyspnoea and are a valuable adjunct to clinical management of COVID-19 associated lung disease. Artificial intelligence (AI) has the potential to facilitate rapid triage of CXRs for further patient testing and/or isolation. In this work we develop an AI algorithm, CovIx, to differentiate normal, abnormal, non-COVID-19 pneumonia, and COVID-19 CXRs using a multicentre cohort of 293,143 CXRs. The algorithm is prospectively validated in 3289 CXRs acquired from patients presenting to ED with symptoms of COVID-19 across four sites in NHS Greater Glasgow and Clyde. CovIx achieves area under receiver operating characteristic curve for COVID-19 of 0.86, with sensitivity and F1-score up to 0.83 and 0.71 respectively, and performs on-par with four board-certified radiologists. AI-based algorithms can identify CXRs with COVID-19 associated pneumonia, as well as distinguish non-COVID pneumonias in symptomatic patients presenting to ED. Pre-trained models and inference scripts are freely available at https://github.com/beringresearch/bravecx-covid.

SUBMITTER: Drozdov I

PROVIDER: S-EPMC8516957 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundThe prevalence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection among asymptomatic patients admitted to hospital has implications for personal protective equipment use, testing strategy and confidence in the safety of acute care services. Our aim was to estimate the positivity rate of reverse transcription polymerase chain reaction (RT-PCR) testing among people admitted to hospital without symptoms of coronavirus disease 2019 (COVID-19) in Alberta, Canada.MethodsBetween Apr. 9 and May 24, 2020, we screened for COVID-19 symptoms and tested for SARS-CoV-2 infection in all consecutive adult patients (≥ 18 yr) admitted via emergency department to 3 Alberta hospitals. We summarized the parameters of the epidemic curve and assessed the performance of symptom screening versus RT-PCR results on nasopharyngeal or oropharyngeal swab samples.ResultsThe study period encompassed Alberta's initial epidemic curve, with peak active cases per 100 000 of 71.4 (0.07%) on Apr. 30, 2020, and 14.7 and 14.6 at the beginning (Apr. 9, 2020) and end (May 24, 2020), respectively. Testing for SARS-CoV-2 infection (64.9% throat and 35.1% nasopharyngeal swabs) was done on 3375 adults (mean age 51, standard deviation 21, yr; 51.5% men). None of the asymptomatic patients (n = 1814) tested positive, and 71 of those with symptoms tested positive (n = 1561; 4.5%, 95% confidence interval [CI] 3.6%-5.7%). Sensitivity of symptom screening (v. RT-PCR) was 100% (95% CI 95%-100%), and specificity was 55% (95% CI 53%-57%). Posttest probabilities for prevalence of SARS-CoV-2 infection ranging from 1.5 to 14 times the peak prevalence of active cases during the study did not change when we assumed lower sensitivity (92%).InterpretationIn a region with low disease prevalence where protocolized symptom assessment was in place during the admission process, we did not identify people admitted to hospital without COVID-19 symptoms who were RT-PCR positive. There may not be additive benefit to universal testing of asymptomatic patients on hospital admission in a setting of low pretest probability and strong public health containment.

Project description:BackgroundHospital emergency departments play a crucial role in the initial assessment and management of suspected COVID-19 infection. This needs to be guided by studies of people presenting with suspected COVID-19, including those admitted and discharged, and those who do not ultimately have COVID-19 confirmed. We aimed to characterise patients attending emergency departments with suspected COVID-19, including subgroups based on sex, ethnicity and COVID-19 test results.Methods and findingsWe 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. Outcomes were admission to hospital, COVID-19 result, organ support (respiratory, cardiovascular or renal), and death, by record review at 30 days. Mean age was 58.4 years, 11200 (50.4%) were female and 11034 (49.6%) male. Adults (age >16 years) were acutely unwell (median NEWS2 score of 4), frequently had limited performance status (46.9%) and had high rates of admission (67.1%), COVID-19 positivity (31.2%), organ support (9.8%) and death (15.5%). Children had much lower rates of admission (27.4%), COVID-19 positivity (1.2%), organ support (1.4%) and death (0.3%). Similar numbers of men and women presented to the ED, but men were more likely to be admitted (72.9% v 61.4%), require organ support (12.2% v 7.7%) and die (18.2% v 13.0%). Black or Asian adults tended to be younger than White adults (median age 54, 50 and 67 years), were less likely to have impaired performance status (43.1%, 26.8% and 51.6%), be admitted to hospital (60.8%, 57.3%, 69.6%) or die (11.6%, 11.2%, 16.4%), 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 suspected and confirmed COVID-19 had similar age, performance status 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 (32.1% v 15.5%).ConclusionsImportant differences exist between patient groups presenting to the emergency department with suspected COVID-19. Adults and children differ markedly and require different approaches to emergency triage. Admission and adverse outcome rates among adults suggest that policies to avoid unnecessary ED attendance achieved their aim. Subsequent COVID-19 confirmation confers a worse prognosis and greater need for organ support.RegistrationISRCTN registry, ISRCTN56149622, http://www.isrctn.com/ISRCTN28342533.

Project description:BackgroundThe health crisis resulting from the global COVID-19 pandemic highlighted more than ever the need for rapid, reliable and safe methods of diagnosis and monitoring of respiratory diseases. To study pulmonary involvement in detail, one of the most common resources is the use of different lung imaging modalities (like chest radiography) to explore the possible affected areas.MethodsThe study of patient characteristics like sex and age in pathologies of this type is crucial for gaining knowledge of the disease and for avoiding biases due to the clear scarcity of data when developing representative systems. In this work, we performed an analysis of these factors in chest X-ray images to identify biases. Specifically, 11 imbalance scenarios were defined with female and male COVID-19 patients present in different proportions for the sex analysis, and 6 scenarios where only one specific age range was used for training for the age factor. In each study, 3 different approaches for automatic COVID-19 screening were used: Normal vs COVID-19, Pneumonia vs COVID-19 and Non-COVID-19 vs COVID-19. The study was validated using two public chest X-ray datasets, allowing a reliable analysis to support the clinical decision-making process.ResultsThe results for the sex-related analysis indicate this factor slightly affects the system in the Normal VS COVID-19 and Pneumonia VS COVID-19 approaches, although the identified differences are not relevant enough to worsen considerably the system. Regarding the age-related analysis, this factor was observed to be influencing the system in a more consistent way than the sex factor, as it was present in all considered scenarios. However, this worsening does not represent a major factor, as it is not of great magnitude.ConclusionsMultiple studies have been conducted in other fields in order to determine if certain patient characteristics such as sex or age influenced these deep learning systems. However, to the best of our knowledge, this study has not been done for COVID-19 despite the urgency and lack of COVID-19 chest x-ray images. The presented results evidenced that the proposed methodology and tested approaches allow a robust and reliable analysis to support the clinical decision-making process in this pandemic scenario.

Dataset Information

Development and prospective validation of COVID-19 chest X-ray screening model for patients attending emergency departments

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets