ABSTRACT
BACKGROUND AND OBJECTIVE: Research on automatic auscultation diagnosis of COVID-19 has not yet been developed. We therefore aimed to engineer a deep learning approach for the automated grading diagnosis of COVID-19 by pulmonary auscultation analysis. METHODS: 172 confirmed cases of COVID-19 in Tongji Hospital were divided into moderate, severe and critical group. Pulmonary auscultation were recorded in 6-10 sites per patient through 3M littmann stethoscope and the data were transferred to computer to construct the dataset. Convolutional neural network (CNN) were designed to generate classifications of the auscultation. F1 score, the area under the curve (AUC) of the receiver operating characteristic curve, sensitivity and specificity were quantified. Another 45 normal patients were served as control group. RESULTS: There are about 56.52%, 59.46% and 78.85% abnormal auscultation in the moderate, severe and critical groups respectively. The model showed promising performance with an averaged F1 scores (0.9938 95% CI 0.9923-0.9952), AUC ROC score (0.9999 95% CI 0.9998-1.0000), sensitivity (0.9938 95% CI 0.9910-0.9965) and specificity (0.9979 95% CI 0.9970-0.9988) in identifying the COVID-19 patients among normal, moderate, severe and critical group. It is capable in identifying crackles, wheezes, phlegm sounds with an averaged F1 scores (0.9475 95% CI 0.9440-0.9508), AUC ROC score (0.9762 95% CI 0.9848-0.9865), sensitivity (0.9482 95% CI 0.9393-0.9578) and specificity (0.9835 95% CI 0.9806-0.9863). CONCLUSIONS: Our model is accurate and efficient in automatically diagnosing COVID-19 according to different categories, laying a promising foundation for AI-enabled auscultation diagnosing systems for lung diseases in clinical applications.
Subject(s)
COVID-19 , Algorithms , Artificial Intelligence , Auscultation , Cohort Studies , Humans , ROC Curve , SARS-CoV-2ABSTRACT
Background and Objective Research on automatic auscultation diagnosis of COVID-19 has not yet been developed. We therefore aimed to engineer a deep learning approach for the automated grading diagnosis of COVID-19 by pulmonary auscultation analysis. Methods 172 confirmed cases of COVID-19 in Tongji Hospital were divided into moderate, severe and critical group. Pulmonary auscultation were recorded in 6-10 sites per patient through 3M littmann stethoscope and the data were transferred to computer to construct the dataset. Convolutional neural network (CNN) were designed to generate classifications of the auscultation. F1 score, the area under the curve (AUC) of the receiver operating characteristic curve, sensitivity and specificity were quantified. Another 45 normal patients were served as control group. Results There are about 56.52%, 59.46% and 78.85% abnormal auscultation in the moderate, severe and critical groups respectively. The model showed promising performance with an averaged F1 scores (0.9938 95% CI 0.9923–0.9952), AUC ROC score (0.9999 95% CI 0.9998–1.0000), sensitivity (0.9938 95% CI 0.9910–0.9965) and specificity (0.9979 95% CI 0.9970–0.9988) in identifying the COVID-19 patients among normal, moderate, severe and critical group. It is capable in identifying crackles, wheezes, phlegm sounds with an averaged F1 scores (0.9475 95% CI 0.9440–0.9508), AUC ROC score (0.9762 95% CI 0.9848–0.9865), sensitivity (0.9482 95% CI 0.9393–0.9578) and specificity (0.9835 95% CI 0.9806–0.9863). Conclusions Our model is accurate and efficient in automatically diagnosing COVID-19 according to different categories, laying a promising foundation for AI-enabled auscultation diagnosing systems for lung diseases in clinical applications.
ABSTRACT
AIMS: In patients with coronavirus disease 2019 (COVID-19), the involvement of the cardiovascular system significantly relates to poor prognosis. However, the risk factors for acute myocardial injury have not been sufficiently studied. Thus, we aimed to determine the characteristics of myocardial injury and define the association between routine blood markers and cardiac troponin I, in order to perform a predictive model. METHODS AND RESULTS: This retrospective cohort study included patients with confirmed COVID-19 from Wuhan Tongji Hospital (Wuhan, China). Data were compared between those with and without myocardial injury. Kaplan-Meier analysis and Cox regression models were used to describe the association between myocardial injury and poor prognosis. Simple correlation analyses were used to find factors associated with high-sensitivity cardiac troponin I levels. Univariate and multivariate logistic regression methods were used to explore the risk factors associated with myocardial injury. The area under the receiver operating characteristic curve was used to determine the predictive value of the model. Of 353 patients included in the study, 79 presented myocardial injury. Patients with myocardial injury had higher levels of inflammation markers, poorer liver and kidney function, and more complications compared with patients without myocardial injury. High-sensitivity cardiac troponin I levels were significantly associated with neutrophil/lymphocyte ratio, creatinine, d-dimer, lactate dehydrogenase, and inflammatory cytokines and negatively associated with oxygen saturation. It was significantly associated with poor prognosis after adjusting for age, sex, and complications. Multivariate regression showed that myocardial injury was associated with a high neutrophil/lymphocyte ratio (odds ratio 2.30, 95% CI 1.11-4.75, per standard deviation increase, P = 0.02), creatinine (3.58, 1.35-8.06, P = 0.01), and lactate dehydrogenase (3.39, 1.42-8.06, P = 0.01) levels. Using a predictive model, the area under the receiver operating characteristic curve was 0.92 (0.88-0.96). CONCLUSIONS: In patients with COVID-19, neutrophil/lymphocyte ratio, creatinine, and lactate dehydrogenase are blood markers that could help identify patients with a high risk of myocardial injury at an early stage.
ABSTRACT
BACKGROUND: To investigate the effect of cardiovascular disease (CVD) on the global pandemic, coronavirus disease 2019 (COVID-19), we analyzed the cases of laboratory-confirmed COVID-19 patients in Wuhan.MethodsâandâResults:Data were extracted from the medical records. SARS-CoV-2 RNA was confirmed by RT-PCR. A total of 33 (53.2%) of 62 cases with CVD, who had higher prevalence of severe COVID-19 compared with non-CVD patients (P=0.027). The median age of all patients was 66.0 (53.3, 73.0) years old. Coronary artery disease (11.3%) and hypertension (38.7%) were the common coexisting CVDs in COVID-19 patients. High-sensitivity cardiac troponin I (hs-cTnI), creatinine, high-density lipoprotein-cholesterol, interleukin-6, C-reactive protein, prothrombin time, and D-dimer levels in the severe COVID-19 with CVD group were higher than in the non-severe COVID-19 with CVD group (P<0.05). For all patients, chest computed tomography (CT) showed ground-glass opacity (66.1%), local (21.0%), bilateral (77.4%), and interstitial abnormalities (4.8%). In COVID-19 patients with CVD, 27 (81.8%) were cured and discharged. 6 (18.2%) remained in hospital, including 2 (3.2%) patients requiring intubation and mechanical ventilation. The hs-cTnI levels in the remaining hospitalized patients were higher than in the discharged patients (P=0.047). CONCLUSIONS: CVDs play a vital role in the disease severity of COVID-19. COVID-19 could result in myocardial injury, which affects the prognosis of COVID-19.