Your browser doesn't support javascript.
Show: 20 | 50 | 100
Results 1 - 4 de 4
Biomedical Signal Processing and Control ; 72:103304, 2022.
Article in English | ScienceDirect | ID: covidwho-1509612


Automatic cough detection in the patients’ realistic audio recordings is of great significance to diagnose and monitor respiratory diseases, such as COVID-19. Many detection methods have been developed so far, but they are still unable to meet the practical requirements. In this paper, we present a deep convolutional bidirectional long short-term memory (C-BiLSTM) model with boundary regression for cough detection, where cough and non-cough parts need to be classified and located. We added convolutional layers before the LSTM to enhance the cough features and preserve the temporal information of the audio data. Considering the importance of the cough event integrity for subsequent analysis, the novel model includes an embedded boundary regression on the last feature map for both higher detection accuracy and more accurate boundaries. We delicately designed, collected and labelled a realistic audio dataset containing recordings of patients with respiratory diseases, named the Corp Dataset. 168 h of recordings with 9969 coughs from 42 different patients are included. The dataset is published online on the MARI Lab website ( The results show that the system achieves a sensitivity of 84.13%, a specificity of 99.82% and an intersection-over-union (IoU) of 0.89, which is significantly superior to other related models. With the proposed method, all the criteria on cough detection significantly increased. The open source Corp Dataset provides useful material and a benchmark for researchers investigating cough detection. We propose the state-of-the-art system with boundary regression, laying the foundation for identifying cough sounds in real-world audio data.

Clinical eHealth ; 3:7-15, 2020.
Article in English | PMC | ID: covidwho-822402


The aim is to diagnose COVID-19 earlier and to improve its treatment by applying medical technology, the “COVID-19 Intelligent Diagnosis and Treatment Assistant Program (nCapp)” based on the Internet of Things. Terminal eight functions can be implemented in real-time online communication with the “cloud” through the page selection key. According to existing data, questionnaires, and check results, the diagnosis is automatically generated as confirmed, suspected, or suspicious of 2019 novel coronavirus (2019-nCoV) infection. It classifies patients into mild, moderate, severe or critical pneumonia. nCapp can also establish an online COVID-19 real-time update database, and it updates the model of diagnosis in real time based on the latest real-world case data to improve diagnostic accuracy. Additionally, nCapp can guide treatment. Front-line physicians, experts, and managers are linked to perform consultation and prevention. nCapp also contributes to the long-term follow-up of patients with COVID-19. The ultimate goal is to enable different levels of COVID-19 diagnosis and treatment among different doctors from different hospitals to upgrade to the national and international through the intelligent assistance of the nCapp system. In this way, we can block disease transmission, avoid physician infection, and epidemic prevention and control as soon as possible.

Lancet Digit Health ; 2(6): e323-e330, 2020 06.
Article in English | MEDLINE | ID: covidwho-260619


Background: The outbreak of COVID-19 has led to international concern. We aimed to establish an effective screening strategy in Shanghai, China, to aid early identification of patients with COVID-19. Methods: We did a multicentre, observational cohort study in fever clinics of 25 hospitals in 16 districts of Shanghai. All patients visiting the clinics within the study period were included. A strategy for COVID-19 screening was presented and then suspected cases were monitored and analysed until they were confirmed as cases or excluded. Logistic regression was used to determine the risk factors of COVID-19. Findings: We enrolled patients visiting fever clinics from Jan 17 to Feb 16, 2020. Among 53 617 patients visiting fever clinics, 1004 (1·9%) were considered as suspected cases, with 188 (0·4% of all patients, 18·7% of suspected cases) eventually diagnosed as confirmed cases. 154 patients with missing data were excluded from the analysis. Exposure history (odds ratio [OR] 4·16, 95% CI 2·74-6·33; p<0·0001), fatigue (OR 1·56, 1·01-2·41; p=0·043), white blood cell count less than 4 × 109 per L (OR 2·44, 1·28-4·64; p=0·0066), lymphocyte count less than 0·8 × 109 per L (OR 1·82, 1·00-3·31; p=0·049), ground glass opacity (OR 1·95, 1·32-2·89; p=0·0009), and having both lungs affected (OR 1·54, 1·04-2·28; p=0·032) were independent risk factors for confirmed COVID-19. Interpretation: The screening strategy was effective for confirming or excluding COVID-19 during the spread of this contagious disease. Relevant independent risk factors identified in this study might be helpful for early recognition of the disease. Funding: National Natural Science Foundation of China.

COVID-19/diagnosis , Adolescent , Adult , Age Factors , Aged , Aged, 80 and over , COVID-19/epidemiology , COVID-19/etiology , COVID-19/pathology , Child , Child, Preschool , China/epidemiology , Female , Fever/etiology , Humans , Infant , Infant, Newborn , Leukocyte Count , Lung/pathology , Male , Middle Aged , Multivariate Analysis , Risk Factors , Young Adult