Data Mining Techniques for Extraction and Analysis of Covid-19 Data

Artificial intelligence has played a crucial role in medical disease diagnosis. In this research, data mining techniques that included deep learning with different scenarios are presented for extraction and analysis of covid-19 data. The energy of the features is implemented and calculated from the CT scan images. A modified meta-heuristic algorithm is introduced and then used in the suggested way to determine the best and most useful features, which are based on how ants behave. Different patients with different problems are investigated and analyzed. Also, the results are compared with other studies. The results of the proposed method show that the proposed method has higher accuracy than other methods. It is concluded from the results that the most crucial features can be concentrated on during feature selection, which lowers the error rate when separating sick from healthy individuals. © 2022 IEEE.

Accuration of Classification of Covid with Convolutional Neural Network-Based Image Chest X-ray with Variations in Image Size and Batch Size

The use of convolutional neural networks in Covid classification has a positive impact on the speed of justification and can provide high accuracy. But on the one hand, the many parameters on CNN will also have an impact on the resulting accuracy. CNN requires time and a heavy level of computation. Setting the right parameters will provide high accuracy. This study examines the performance of CNN with variations in image size and minibatch. Parameter settings used are max epoch values of 100, minibatch variations of 32, 64, and 128, and learning rate of 0.1 with image size inputs of 50,100, and 150 variations on the level of accuracy. The dataset consists of training data and test data, 200 images, which are divided into two categories of normal and abnormal images (Covid). The results showed an accuracy with the use of minibatch 128 with the highest level of accuracy at image size 150 × 150 on test data of 99,08%. The size of the input matrix does not always have an impact on increasing the level of accuracy, especially on the minibatch 32. The parameter setting on CNN was dependent on the CNN architecture, the dataset used, and the size of the dataset. One can imply that optimization parameter in CNN can approve good accuration. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

Detection of Pneumonia by Binary Image Classification Using Hybrid Neural Networks

Pneumonia is a more contagious virus with worldwide health implications. If positive cases are detected early enough, spread of the pandemic sickness can be slowed. Pneumonia illness estimation is useful for identifying patients who are at risk of developing health problems. So, the conventional method like PCR kits used to detect the covid patients lead to an increase in pneumonia cases as it failed to detect at the earliest. A polymerase chain reaction (PCR) test will be performed right away on the blood or sputum to quickly identify the DNA of the bacteria that cause pneumonia. With the help of CXR images, the pneumonia is diagnosed with a high accuracy rate utilizing the HNN (Hybrid Neural Network) method. Thus, isolating them at the earlier stage and preventing the spread of disease. © 2022 IEEE.

Towards Explainable AI on Chest X-Ray Diagnosis Using Image Segmentation and CAM Visualization

Computer vision in medical diagnosis has achieved a high level of success in diagnosing diseases with high accuracy. However, conventional classifiers that produce an image-to-label result provide insufficient information for medical professionals to judge and raise concerns over the trust and reliability of a model with results that cannot be explained. Class activation maps are a method of providing insight into a convolutional neural network's feature maps that lead to its classification but in the case of lung diseases, the region of concern is only the lungs. Therefore, the proposed model combines image segmentation models and classifiers to crop out only the lung region of a chest X-ray's class activation map to provide a visualization that improves the explainability and trust of an AI's diagnosis by focusing on a model's weights within the region of concern. The proposed U-Net model achieves 97.72% accuracy and a dice coefficient of 0.9691 on testing data from the COVID-QU-Ex Dataset which includes both diseased and healthy lungs. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Face Mask Detection and Recognition with High Accuracy on Live Streaming Video Using Improved Yolo V4 and Comparing with Convolutional Neural Network

The aim of this research is to detect face masks using Convolutional Neural network (CNN) algorithm and comparing it with the Yolo v4 algorithm. The study includes two groups namely, CNN algorithm and yolo v4 algorithm. The total sample size is 40 with pretest power of 0.8. In order to evaluate how well CNN algorithm methods perform, accuracy values are calculated. Using SPSS software, CNN algorithm method was found to be 92.65% accurate while improved Yolo v4 was found to be 85.87% accurate. 0.000 p(2-tailed) is obtained for the model. Using CNN, it was proved significant improvements to performance than improved Yolo v4. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Face mask detection using transfer learning and OpenCV in live videos

In 2019, we have seen the biggest epidemic of the century, which claimed many lives worldwide. The epidemic has in fact changed our life in many ways. It changed the way we interact with people. Wearing a mask is now the new normal. Though now the vaccine for the disease is available, still wearing a mask can save us from Covid19, its variants, and other contagious diseases.Especially at places where the large gathering is expected wearing a mask can be made mandatory and our proposed framework can do its monitoring through CCTV cameras.So in this research, we build a deep learning-based framework to detect whether some person is wearing a mask or not through the live video stream. We used a total of three state-of-The-Art transfer learning methods to train our system and used OpenCV to detect faces in the live video stream. We found that efficientnetB1 achieved the highest accuracy of 97.75%. © 2022 IEEE.

Development of Website for COVID-19 Detection on Chest X-Ray Images

Control of the spread of COVID-19 must be encouraged, even though this is a new normal era. Rapid screening for COVID-19 detection must be carried out to control the spread of COVID-19. This research develops a website for COVID-19 detection based on chest X-Ray images and compares the CNN-BiLSTM model. This study divides X-ray images of the chest into three categories: COVID-19, Normal, and Viral Pneumonia. When compared to other models, the Resnet50-BiLSTM model produces the highest accuracy. The accuracy of the Resnet50-BiLSTM model was 98.51%. Then, in order, the following models were used: Resnet50, VGG19-BiLSTM, VGG19, AlexNet-BiLSTM, and AlexNet. The comparison of Precision, Recall, and F1-Measure findings also demonstrate that Resnet50-BiLSTM has the highest score when compared to other approaches. The website was also developed using the Flask framework for automatic COVID-19 detection. © 2022 IEEE.

Social Distance Alerting Device for COVID-19

We are proposing an IoT-based social distancing device as a preventive measure to COVID-19. It uses NodeMCU in conjunction with ultrasonic sensor temperature sensor, while vibrator buzzer is used for an alarming mechanism. The ultrasonic sensor is used to obtain higher accuracy as it uses LOS principle to measure the distance. The alarm will be raised whenever measured distance is found to be less than six feet. Temperature sensor is used to alert the user to isolate them if their body temperature goes above 102 °F, thereby decreasing the transmission possibility of virus in case he is infected with the virus. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

Exploring Generalizability of Fine-Tuned Models for Fake News Detection

The Covid-19 pandemic has caused a dramatic and parallel rise in dangerous misinformation, denoted an 'infodemic' by the CDC and WHO. Misinformation tied to the Covid-19 infodemic changes continuously;this can lead to performance degradation of fine-tuned models due to concept drift. Degredation can be mitigated if models generalize well-enough to capture some cyclical aspects of drifted data. In this paper, we explore generalizability of pre-trained and fine-tuned fake news detectors across 9 fake news datasets. We show that existing models often overfit on their training dataset and have poor performance on unseen data. However, on some subsets of unseen data that overlap with training data, models have higher accuracy. Based on this observation, we also present KMeans-Proxy, a fast and effective method based on K-Means clustering for quickly identifying these overlapping subsets of unseen data. KMeans-Proxy improves generalizability on unseen fake news datasets by 0.1-0.2 f1-points across datasets. We present both our generalizability experiments as well as KMeans-Proxy to further research in tackling the fake news problem. © 2022 IEEE.

iConDet2: An Improved Conjunctivitis Detection Portable Healthcare App Powered by Artificial Intelligence

Conjunctivitis is one of the common and contagious ocular diseases which affects the conjunctiva of the human eye. Both the bacterial and viral types of it can be treated with eye drops and other medicines. It is important to diagnose the disease at its early stage to realise the connection between it and other diseases, especially COVID-19. Mobile applications like iConDet is such a solution that performs well for the initial screening of Conjunctivitis. In this work, we present with iConDet2 which provides an advanced solution than the earlier version of it. It is faster with a higher accuracy level (95%) than the previously released iConDet. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

COVIDGuardian: A Machine Learning approach for detecting the Three Cs

On January 30, 2020, WHO officially declared the outbreak of COVID-19 a Public Health Emergency of International Concern. Japan announced the state of emergency and implemented safety protocols the "Three Cs", a warning guideline addressing to voluntarily avoid potentially COVID-19 hazardous situations such as confined and closed spaces, crowded places and close-contact settings that lead to occurrence of serious clusters. The primary goal of this research is to identify the factors which help to estimate whether the user is in the Three Cs. We propose COVIDGuardian, a system that detects the Three Cs based on data such as CO2, temperature, humidity, and wireless packet log. The results show that estimation of closed space had the highest accuracy followed by close-contact settings and crowded places. The ensemble Random Forest (RF) classifier demonstrates the highest accuracy and F score in detecting closed spaces and crowded spaces. The findings indicated that integrated loudness value, average CO2, average humidity, probe request log, and average RSSI are of critical importance. In addition, when the probe request logs were filtered at three RSSI cutoff points (1m, 3m, and 5m), 1m cut-off points had the highest accuracy and F Score among the Three C models. © 2022 Copyright held by the owner/author(s).

An intelligent guided-airborne virus detection

Health has recently faced many challenges, including improving a healthy environment and reducing human life's dangers and economic crises. The last pandemic COVID-19 had badly affected survivor sectors with infection and lockdown exigence. Scientists proposed several solutions to reduce the negative impact of a such pandemic by proposing systems for earlier detection of viruses. The use of metamaterials as an emerging technology in the biosensors field allows a high accuracy. This paper presents a method for detecting and capturing airborne viruses using metasurface technology. The goal is to develop a system that can identify and capture these viruses using FET sensors. The accuracy of the detection is tied to the concentration of aerosols. The model proposes a guided flow of aerosols that positively impacts the detection of viruses through the FET biosensor. The simulation results based on Concentration and airflow velocity delays prove the proposed model's performance. © 2022 IEEE.

COVIDGuardian: A Machine Learning approach for detecting the Three Cs

On January 30, 2020, WHO officially declared the outbreak of COVID-19 a Public Health Emergency of International Concern. Japan announced the state of emergency and implemented safety protocols the "Three Cs", a warning guideline addressing to voluntarily avoid potentially COVID-19 hazardous situations such as confined and closed spaces, crowded places and close-contact settings that lead to occurrence of serious clusters. The primary goal of this research is to identify the factors which help to estimate whether the user is in the Three Cs. We propose COVIDGuardian, a system that detects the Three Cs based on data such as CO2, temperature, humidity, and wireless packet log. The results show that estimation of closed space had the highest accuracy followed by close-contact settings and crowded places. The ensemble Random Forest (RF) classifier demonstrates the highest accuracy and F score in detecting closed spaces and crowded spaces. The findings indicated that integrated loudness value, average CO2, average humidity, probe request log, and average RSSI are of critical importance. In addition, when the probe request logs were filtered at three RSSI cutoff points (1m, 3m, and 5m), 1m cut-off points had the highest accuracy and F Score among the Three C models. © 2022 Copyright held by the owner/author(s).

Fine-Tuning BERT for Question and Answering Using PubMed Abstract Dataset

The coronavirus, which first originated in China in 2019, spread worldwide and eventually reached a pandemic situation. In the interest of many people, misinformation about the coronavirus has been pouring out on the Internet. We developed a Q&A processing technique by building a dataset based on the PubMed paper for people to easily get the right information. We fine-tuned BioBERT among the BERT models that reached SOTA performance in the biomedical Q&A task. It answered questions about coronavirus with high accuracy. In the future, we will develop our technology that can handle Q&A not only in English but also in multiple languages. This work will contribute to helping people who speak different languages easily obtain correct information amidst confusing data. © 2022 Asia-Pacific of Signal and Information Processing Association (APSIPA).

Visual Servo Control of COVID-19 Nasopharyngeal Swab Sampling Robot

In this study, we present a visual servo control framework for fully automated nasopharyngeal swab robots. The proposed framework incorporates a deep learning-based nostril detection with a cascade approach to reliably identify the nostrils with high accuracy in real time. In addition, a partitioned visual servoing scheme that combines image-based visual servoing with axial control is formulated for accurately positioning the sampling swabs at the nostril with a multi-DOF robot arm. As the visual servoing is designed to minimize an error between the detected nostril and the swab, it can compensate for potential errors in real operation, such as positioning error by inaccurate camera-robot calibration and kinematic error by unavoidable swab deflection. The performance of the visual servo control was tested on a head phantom model for 30 unused swabs, and then compared with a method referring to only the 3D nostril target for control. Consequently, the swabs reached the nostril target with less than an average error of 1.2±0.5 mm and a maximum error of 2.0 mm via the visual servo control, while the operation without visual feedback yielded an average error of 10.6±2.3 mm and a maximum error of 16.2 mm. The partitioned visual servoing allows the swab to rapidly converge to the nostril target within 1.0 s without control instability. Finally, the swab placement at the nostril among the entire procedure of fully automated NP swab was successfully demonstrated on a human subject via the visual servo control. © 2022 IEEE.

Itinerary Card Anti-cheating Beyond OCR

Since the COVID-19 pandemic, the itinerary card has become so pertinent to our lives that we need to show our itinerary card whether we take public transport or enter public places. The traditional way to manually check and record the information on the itinerary card is inefficient. It easily leads to congestion at the entrance, especially in high-traffic areas. In addition, some people even falsify their itinerary codes to evade mandatory testing or quarantine for COVID-19. Therefore, an efficient itinerary checking method is needed to alleviate the crowded problem, reduce cross-infection, and intelligently detect itinerary cheating. To address these issues, we propose a deep-learning-based method combined with OCR techniques. This method consists of five parts, including ROI locating, color classification, OCR, information pooling, and anti-cheating. The proposed scheme can extract the itinerary information on the itinerary card and check it. It also provides a certain anti-cheating function. Experimental results show that the proposed scheme can efficiently check the information on the itinerary card with high accuracy. © 2022 IEEE.

Detection of Mask-wearing Based on Improved YOLOV3

To control the spread of the virus, mask detection is crucial in public areas, especially after the outbreak of Covid-19 pneumonia. This paper aims to improve the accuracy and precision of mask detection. This study improves mask-wearing detection by adding data augmentation, using the smooth label to replace the one-hot vector, and customizing the network connection of the YOLOv3 network. Through these targeted improvements, the average precision of face with mask detection has been increased by 0.9%, and the average precision of face without mask detection has been increased by 2.9%, which implies that it is a better strategy to do mask detection based on YOLOv3. By inputting photographs, the network can check, with high accuracy, whether the pedestrians in the picture wear masks or not, which will be a good supplementary to epidemic prevention and control. © 2022 SPIE.

Set-Valued Medical Image Classification with Evidential CNN: A First Test with Covid-19 Dataset

Deep learning based models have been achieving ever high accuracy for precise image classification. However, in the medical sector where decisions should to be made more cautiously and where inaccuracy is less of a concern than precision and recall, it might be more appropriate to resort to imprecise classifiers, e.g. set-valued classifiers. In this work, an evidential convolutional neural network (ECNN) method is applied for set-valued medical image classification with Covid-19 X-ray dataset. Experimental result shows that the ECNN classifier is able to assign confusing image patterns to multi-class sets, while maintaining high accuracy compared to traditional probabilistic CNN classifiers. This result reveals that the ECNN classifier holds good promise of being applied for imprecise medical image classification. © 2022 IEEE.

Custom Weighted Balanced Loss function for Covid 19 Detection from an Imbalanced CXR Dataset

In this paper, we have proposed a novel framework, that is ResNet-18 model along with Custom Weighted Balanced loss function, in order to automatically detect Covid-19 disease from a highly imbalanced Chest X-Ray (CXR) dataset. Covid 19 disease has become a global pandemic, for last two years. Early automatic detection of Covid-19, from CXR images has been the key to survive from this pandemic. In the recent advent, researchers have already proposed several Deep Learning (DL) models, which can detect Covid-19 disease (with higher accuracy) from CXR images. However, Covid-19 detection by DL models are fraught with the problem of class imbalance, since most of the available CXR datasets are found highly imbalanced. In this paper, we have worked in a new direction, that is, alleviating the class imbalance problem from CXR dataset by using novel loss function. First, we choose a challengeable CXR dataset in which there are four classes, they are Covid, Normal, Lung Opacity (LO) and Viral Pneumonia (VP). Later we have identified that real problem of this dataset is not only the class imbalance, but also, huge intra-class variance is observed in Covid class. Therefore, we have come up with a new idea, that is, modifying the bias weights in a Weighted Categorical Cross Entropy (WCCE), based on reducing both of the factors, i.e., class imbalance and intra-class variance from the dataset. For the experimentation, we have chosen a ResNet-18 model which is trained from scratch for a large Chexpert CXR dataset and thereafter it is pre-trained on the Covid CXR dataset. Experimental results suggest that ResNet-18 model along with proposed Custom Weighted Balanced loss function, have improved 2-4% accuracy, precision, recall, F1 score and AUC for four class CXR dataset. Furthermore, we have tested the same framework for three class Covid CXR dataset, after excluding LO class. We have achieved 96% accuracy, 97% precision, 96% recall, 97% F1 score and 97% AUC for three class classification task. This is significant (3-4%) improvement than the performance of ResNet-18 model with CCE. © 2022 IEEE.

Classifying the Students' Behavior on e-Learning System using Fine-Tuning K-NN Method

monitoring the student's behavior is challenging for teachers in online learning, which is crucial to solving. It is because, in this pandemic period, online learning is required to minimize the spreading of coronavirus. However, research in this domain is not much. This study provides an alternative to this problem by classifying students' behavior in the e-Learning system, where the k-NN is applied to mine the students' behavior. In addition, this paper also tests the proper parameters to improve the performance of k-NN: k and distance. The experimental result shows that the best performance on the cross-validation technique is reached by Euclidean distance and, on the percentage-split, is achieved by distance-Manhattan. These are indicated by the highest accuracy level obtained by neighbors of five and 20 fold, about 96.9% on the cross-validation technique. On the percentage split technique, the highest accuracy level, about 95.3%, is reached by neighbors of four and split 50%. In this best performance, four students are misclassified on the cross-validation and six on the percentage split. © 2022 IEEE.