Unsupervised clustering of SARS-CoV-2 using deep convolutional autoencoder

Clustering analysis for the evolutionary relationships of SARS-CoV-2 strains

To explore the differences and relationships between the available SARS-CoV-2 strains and predict the potential evolutionary direction of these strains, we employ the hierarchical clustering analysis to investigate the evolutionary relationships between the SARS-CoV-2 strains utilizing the genomic sequences collected in China till January 7, 2023. We encode the sequences of the existing SARS-CoV-2 strains into numerical data through k-mer algorithm, then propose four methods to select the representative sample from each type of strains to comprise the dataset for clustering analysis. Three hierarchical clustering algorithms named Ward-Euclidean, Ward-Jaccard, and Average-Euclidean are introduced through combing the Euclidean and Jaccard distance with the Ward and Average linkage clustering algorithms embedded in the OriginPro software. Experimental results reveal that BF.28, BE.1.1.1, BA.5.3, and BA.5.6.4 strains exhibit distinct characteristics which are not observed in other types of SARS-CoV-2 strains, suggesting their being the majority potential sources which the future SARS-CoV-2 strains' evolution from. Moreover, BA.2.75, CH.1.1, BA.2, BA.5.1.3, BF.7, and B.1.1.214 strains demonstrate enhanced abilities in terms of immune evasion, transmissibility, and pathogenicity. Hence, closely monitoring the evolutionary trends of these strains is crucial to mitigate their impact on public health and society as far as possible.

Classification of tropical cyclone rain patterns using convolutional autoencoder

Heavy rainfall produced by tropical cyclones (TCs) frequently causes wide-spread damage. TCs have different patterns of rain depending on their development stage, geographical location, and surrounding environmental conditions. However, an objective system for classifying TC rain patterns has not yet been established. This study objectively classifies rain patterns of North Atlantic TCs using a Convolutional Autoencoder (CAE). The CAE is trained with 11,991 images of TC rain rates obtained from satellite precipitation estimates during 2000-2020. The CAE consists of an encoder which compresses the original TC rain image into low-dimensional features and a decoder which reconstructs an image from the compressed features. Then, TC rain images are classified by applying a k-means method to the compressed features from the CAE. We identified six TC rain patterns over the North Atlantic and confirmed that they exhibited unique characteristics in their spatial patterns (e.g., area, asymmetry, dispersion) and geographical locations. Furthermore, the characteristics of rain patterns in each cluster were closely related to storm intensity and surrounding environmental conditions of moisture supply, vertical wind shear, and land interaction. This classification of TC rain patterns and further investigations into their evolution and spatial variability can improve forecasts and help mitigate damage from these systems.

Deep learning techniques for detection and prediction of pandemic diseases: a systematic literature review

Deep learning (DL) is becoming a fast-growing field in the medical domain and it helps in the timely detection of any infectious disease (IDs) and is essential to the management of diseases and the prediction of future occurrences. Many scientists and scholars have implemented DL techniques for the detection and prediction of pandemics, IDs and other healthcare-related purposes, these outcomes are with various limitations and research gaps. For the purpose of achieving an accurate, efficient and less complicated DL-based system for the detection and prediction of pandemics, therefore, this study carried out a systematic literature review (SLR) on the detection and prediction of pandemics using DL techniques. The survey is anchored by four objectives and a state-of-the-art review of forty-five papers out of seven hundred and ninety papers retrieved from different scholarly databases was carried out in this study to analyze and evaluate the trend of DL techniques application areas in the detection and prediction of pandemics. This study used various tables and graphs to analyze the extracted related articles from various online scholarly repositories and the analysis showed that DL techniques have a good tool in pandemic detection and prediction. Scopus and Web of Science repositories are given attention in this current because they contain suitable scientific findings in the subject area. Finally, the state-of-the-art review presents forty-four (44) studies of various DL technique performances. The challenges identified from the literature include the low performance of the model due to computational complexities, improper labeling and the absence of a high-quality dataset among others. This survey suggests possible solutions such as the development of improved DL-based techniques or the reduction of the output layer of DL-based architecture for the detection and prediction of pandemic-prone diseases as future considerations.