COVID-19 Hierarchical Classification Using a Deep Learning Multi-Modal.

Coronavirus disease 2019 (COVID-19), originating in China, has rapidly spread worldwide. Physicians must examine infected patients and make timely decisions to isolate them. However, completing these processes is difficult due to limited time and availability of expert radiologists, as well as limitations of the reverse-transcription polymerase chain reaction (RT-PCR) method. Deep learning, a sophisticated machine learning technique, leverages radiological imaging modalities for disease diagnosis and image classification tasks. Previous research on COVID-19 classification has encountered several limitations, including binary classification methods, single-feature modalities, small public datasets, and reliance on CT diagnostic processes. Additionally, studies have often utilized a flat structure, disregarding the hierarchical structure of pneumonia classification. This study aims to overcome these limitations by identifying pneumonia caused by COVID-19, distinguishing it from other types of pneumonia and healthy lungs using chest X-ray (CXR) images and related tabular medical data, and demonstrate the value of incorporating tabular medical data in achieving more accurate diagnoses. Resnet-based and VGG-based pre-trained convolutional neural network (CNN) models were employed to extract features, which were then combined using early fusion for the classification of eight distinct classes. We leveraged the hierarchal structure of pneumonia classification within our approach to achieve improved classification outcomes. Since an imbalanced dataset is common in this field, a variety of versions of generative adversarial networks (GANs) were used to generate synthetic data. The proposed approach tested in our private datasets of 4523 patients achieved a macro-avg F1-score of 95.9% and an F1-score of 87.5% for COVID-19 identification using a Resnet-based structure. In conclusion, in this study, we were able to create an accurate deep learning multi-modal to diagnose COVID-19 and differentiate it from other kinds of pneumonia and normal lungs, which will enhance the radiological diagnostic process.

Group Differences in Facial Emotion Expression in Autism: Evidence for the Utility of Machine Classification.

Effective social communication relies, in part, on accurate nonverbal expression of emotion. To evaluate the nature of facial emotion expression (FEE) deficits in children with autism spectrum disorder (ASD), we compared 20 youths with ASD to a sample of typically developing (TD) youth (n = 20) using a machine-based classifier of FEE. Results indicate group differences in FEE for overall accuracy across emotions. In particular, a significant group difference in accuracy of FEE was observed when participants were prompted by a video of a human expressing an emotion, F(2, 36) = 4.99, p = .032, η2 = .12. Specifically, youth with ASD made significantly more errors in FEE relative to TD youth. Findings support continued refinement of machine-based approaches to assess and potentially remediate FEE impairment in youth with ASD.

Feasibility of Automated Training for Facial Emotion Expression and Recognition in Autism.

Impairment in facial emotion recognition (FER) and facial emotion expression (FEE), often documented in autism spectrum disorder (ASD), are believed to contribute to the observed core social-communication disability that characterizes this disorder. Moreover, impaired FER and FEE are frequently seen in other disorders and problem behaviors. We describe the development of a novel system to detect and give real-time feedback on these processes, termed facial emotion expression training (FEET), an automated, gamelike system that is based on 3-dimensional sensing (Kinect) technology. A sample of 40 children (n = 20 ASD, n = 20 typically developing) interacted with our prototype system, which presented audiovisual stimuli and assessed responses of participants. Overall, consumer satisfaction ratings were high, and youth with ASD reported enjoying interacting with the system more than did the typical youth. Results suggest that new technology-based interventions are acceptable to consumers and viable for use in remediation of transdiagnostic processes, such as FER and FEE. Implications for future technology-based intervention to target transdiagnostic processes are discussed.

Nosocomial Clostridium difficile-associated diarrhoea in Assiut University Children's Hospital, Egypt.

BACKGROUND: There are no large epidemiological studies of Clostridium difficile-associated disease (CDAD) in hospitalised children. AIM: To describe the frequency, demography, clinical features and outcome of nosocomial CDAD in children admitted to Assiut University Children's Hospital, Egypt. PATIENTS AND METHODS: In this descriptive cross-sectional study, 72 children developed nosocomial diarrhoea between April 2010 and March 2011. A medical history, clinical assessment and culture for Clostridium difficile and direct toxin detection from stool samples by enzyme immuno-assay were undertaken in all patients. RESULTS: CDAD was diagnosed in 17 (23·6%) patients. Those aged ≤12 months were the most commonly affected (eight, 47%). The main cause of admission was respiratory disorders (eight, 47% of cases), followed by surgical problems (three, 17·7%). Ten patients (58·8%) had severe symptoms. There were no statistically significant differences between any of the demographic or laboratory data for children with CDAD and children with other causes of nosocomial diarrhoea. None of the patients developed complications. Seven children with CDAD (41·2%) had recurrence. CONCLUSION: CDAD is an important cause of nosocomial diarrhoea in children in Assiut University Children's Hospital. Established guidelines should be followed in all hospitals to minimise exposure to the pathogen. Physicians can do much to reduce the risk of a severe outcome in children by early identification and rapid management. Further research should be undertaken to identify the risk factors for recurrence.

Clinico-laboratory study on children with auto-immune hepatitis in Upper Egypt.

BACKGROUND AND STUDY AIMS: Auto-immune hepatitis (AIH) in children is a rare chronic progressive liver disorder. It is characterised serologically by high aminotransferase levels, elevated immunoglobulin G (IgG) and the presence of autoantibodies. AIH is divided into two types according to the autoantibody profile. This study aims to assess frequency, clinical manifestations, biochemical features and outcome of AIH in children attending Assuit University Hospitals in Upper Egypt with acute icteric hepatitis and seronegative viral markers (anti-hepatitis A virus (HAV) IgM, HbsAg, anti-hepatitis C virus (anti-HCV) Ab). PATIENTS AND METHODS: The study includes 34 children with AIH, diagnosed on the basis of the International Scoring Criteria of Auto-immune Hepatitis, recruited from Assuit University Hospitals, during the period from January 2005 to December 2009. All patients received prednisolone 2mgkg(-1)day(-1). Follow-up was done for 1year. RESULTS: Among 34 children diagnosed as AIH, 24 were females (70.5%) and 10 were males (29.5%). Jaundice represented the most consistent finding in all patients. According to the autoantibody profile, 25 children were classified as type 1 and nine children were classified as type 2. Corticosteroid therapy was started. Complete remission was observed in 67.6% of patients and partial remission in 17.6%. There was no significant statistical difference in clinical and biochemical features of AIH in patients regarding the response to treatment. Mild side effects of steroid therapy were encountered in 48.2% of patients. After complete withdrawal of corticosteroids, six patients (20.7%) developed relapse. CONCLUSION: AIH type 1 was the main form of AIH in children referred to Assiut University Hospitals. Girls were more affected than boys. AIH type 1 exhibited a more active, ongoing immunologic process. Steroid alone can be used successfully in most cases. Children with AIH type 2 had a higher frequency of relapse after corticosteroid withdrawal. Further studies on a larger number of cases and long-term follow-up are recommended.