Search

This study presents a comprehensive review of the potential of multimodal deep learning (DL) in medical diagnosis, using COVID-19 as a case example. Motivated by the success of artificial intelligence applications during the COVID-19 pandemic, this research aims to uncover the capabilities of DL in disease screening, prediction, and classification, and to derive insights that enhance the resilience, sustainability, and inclusiveness of science, technology, and innovation systems. Adopting a systematic approach, we investigate the fundamental methodologies, data sources, preprocessing steps, and challenges encountered in various studies and implementations. We explore the architecture of deep learning models, emphasising their data-specific structures and underlying algorithms. Subsequently, we compare different deep learning strategies utilised in COVID-19 analysis, evaluating them based on methodology, data, performance, and prerequisites for future research. By examining diverse data types and diagnostic modalities, this research contributes to scientific understanding and knowledge of the multimodal application of DL and its effectiveness in diagnosis. We have implemented and analysed 11 deep learning models using COVID-19 image, text, and speech (ie, cough) data. Our analysis revealed that the MobileNet model achieved the highest accuracy of 99.97% for COVID-19 image data and 93.73% for speech data (i.e., cough). However, the BiGRU model demonstrated superior performance in COVID-19 text classification with an accuracy of 99.89%. The broader implications of this research suggest potential benefits for other domains and disciplines that could leverage deep learning techniques for image, text, and speech analysis.

As the demand for high-speed data transmission grows with the expected emergence of 6G networks and the proliferation of wireless devices, more than traditional wireless infrastructure may be required. Small cell networks (ScNs) integrated with reconfigurable intelligent surfaces (RISs) and multiple-inputmultiple-output (MIMO) have emerged as promising solutions to address this issue. However, ScNs have resource allocation limitations, and traditional RISs can only reflect signals in a limited propagation space of 1800 with fixed reflection properties. This paper proposes a novel approach to overcome these challenges by introducing actively simultaneously transmitting and reflecting (ASTAR)-RISs. Unlike conventional RIS, ASTAR-RISs actively amplify and transmit signals, effectively mitigating the limited propagation challenge and improving signal strength, especially in dense ScNs. This approach enhances the quality of service in complex channel environments by amplifying, on top of reflection, from the macro base station (mBS), improving the overall signal strength, and providing 3600 flexible propagation space. Furthermore, ASTAR-RIS enables dynamic beam management, significantly improving signal coverage and interference management, which are crucial in dense deployments. In this work, we propose a network architecture where distributed ASTAR-RIS units are deployed to assist small cell mBSs by optimizing signal coverage and enhancing communication performance. ASTAR-RISs dynamically control signal reflection and amplification, complementing the functionality of traditional small-cell BSs in dense network environments. Using the MIMO technique, we design phase shifts for ASTAR elements and develop optimal hybrid beamforming for users at the mBS. We dynamically control the ON/OFF status of the ASTAR-RIS based on active or idle status. We propose an efficient model that ensures fairness of signal-to-noise ratio (SNR) for all users and minimizes overall power consumption while meeting user SNR and phase shift constraints. To this end, we integrate robust beamforming and power allocation strategies, ensuring the system maintains reliable performance even under imperfect channel state information (CSI). We formulate a max-min optimization problem that optimizes the SNR and power consumption, subject to the ON/OFF status, phase shift, and power budget of the ASTAR-RIS. Our proposed method uses an alternating optimization algorithm to optimize the phase shift matrix at the ASTAR-RIS and the hybrid beamforming at the mBS. The approach includes two transmission schemes, and the phase optimization problem is solved using a successive convex approximation method that offers a closed-form solution at each step. Additionally, we use the dual method to determine the optimal ON/OFF status of the ASTAR-RIS. Comprehensive simulations validate the robustness and scalability of our proposed solution, particularly under varying network densities and CSI uncertainties. provides significant performance improvements over 170% compared to traditional RIS schemes.

Abstract Transparent conducting oxides (TCOs) are crucial for high-performance displays, solar cells, and wearable sensors. However, their high process temperatures and brittle nature have hindered their use in flexible electronics. In this paper, we overturn these limitations by harnessing Cabrera-Mott oxidation to fabricate large-area, two-dimensional (2D) transparent electrodes via liquid metal printing. Our robotic, vacuum-free process deposits ultrathin (2–10 nm) indium tin oxide (ITO) with exceptional flexibility, transparency (>95%) and conductivity (>1300 S/cm) by utilizing hypoeutectic In-Sn alloys to print at

Abstract Objectives Serum protein abundance was assessed in adult and juvenile dermatomyositis (DM and JDM) patients to determine differentially regulated proteins, altered pathways, and candidate disease activity biomarkers. Methods Serum protein expression from 17 active adult DM and JDM patients each was compared to matched, healthy control subjects by a multiplex immunoassay. Pathway analysis and protein clustering of the differentially regulated proteins were examined to assess underlying mechanisms. Candidate disease activity biomarkers were identified by correlating protein expression with disease activity measures. Results Seventy-eight of 172 proteins were differentially expressed in the sera of DM and JDM patients compared to healthy controls. Forty-eight proteins were differentially expressed in DM, 32 proteins in JDM, and 14 proteins in both DM and JDM. Twelve additional differentially expressed proteins were identified after combining the DM and JDM cohorts. C-X-C motif chemokine ligand 10 (CXCL10) was the most strongly upregulated protein in both DM and JDM sera. Other highly upregulated proteins in DM included S100 calcium binding protein A12 (S100A12), CXCL9, and nicotinamide phosphoribosyltransferase (NAMPT), while highly upregulated proteins in JDM included matrix metallopeptidase 3 (MMP3), growth differentiation factor 15 (GDF15), and von Willebrand factor (vWF). Pathway analysis indicated that phosphoinositide 3-kinase (PI3K), p38 mitogen-activated protein kinase (MAPK), and toll-like receptor 7 (TLR7) signaling were activated in DM and JDM. Additional pathways specific to DM or JDM were identified. A protein cluster associated with neutrophils and mononuclear leukocytes and a cluster of interferon-associated proteins were observed in both DM and JDM. Twenty-two proteins in DM and 24 proteins in JDM sera correlated with global, muscle, and/or skin disease activity. Seven proteins correlated with disease activity measures in both DM and JDM sera. IL-1 receptor like 1 (IL1RL1) emerged as a candidate global disease activity biomarker in DM and JDM. Conclusion Coordinate analysis of protein expression in DM and JDM patient sera by a multiplex immunoassay validated previous gene expression studies and identified novel dysregulated proteins, altered signaling pathways, and candidate disease activity biomarkers. These findings may further inform the assessment of DM and JDM patients and aid in the identification of potential therapeutic targets.

Traditional techniques have been used in fish farming in Bangladesh since ancient times. However, nowadays the adoption of modern technologies in fish farming in Bangladesh has revolutionized the industry. The goal of this research was to investigate the factors that influence fish farmers' decisions to accept new technologies for aquaculture and to ascertain the degree to which fish farmers have adopted these new technologies. A total of 101 fish farmers in the Trishal Upazila of the Mymensingh district were randomly selected and interviewed using a structured interview schedule. The research found that most of the sample fish farmers were middle to old-aged, had small farm sizes, and earned low to medium income. A significant number of fish farmers didn't receive any training, and about half of them had moderate farming experience. The majority of respondents in this survey had average levels of extended media interaction, organizational engagement, and fish farming expertise. Based on the statistics, we can deduce that 46.53 % of fish farmers made minimal use of some fish farming technology, while 20.79 % used them very little, and 32.67 % utilized them extensively. From the list of eleven socio-demographic characteristics, adopting modern aquaculture technologies was positively and significantly correlated with education, farm size, annual fish farming income, experience, and knowledge of fish farming. The most significant positive link was seen between fish farming experience and adoption. The survey revealed that financial stress, lack of adequate training, market instability, and resource deficits are the major obstacles to modern aquaculture technology adoption, while cultural resistance is the least significant. These factors collectively slow the progress of modernizing fish farming in Bangladesh, despite the potential benefits of improved yields and sustainability. Hence, by resolving these issues, the study area will have more space to employ advanced aquaculture technologies, paving the way to sustainable fish farming in Bangladesh.

Two-dimensional (2D) molybdenum disulfide (MoS2) holds immense promise for next-generation electronic applications. However, the role of contact deposition at the metal/semiconductor interface remains a critical factor influencing device performance. This study investigates the impact of different metal deposition techniques, specifically electron-beam evaporation and sputtering, for depositing Cu, Pd, Bi, Sn, Pt, and In. Utilizing Raman spectroscopy with backside illumination, we observe changes at the buried metal/1L MoS2 interface after metal deposition. Sputter deposition causes more damage to monolayer MoS2 than electron-beam evaporation, as indicated by partial or complete disappearance of first-order E′(Γ)α and A′1(Γ)α Raman modes post-deposition. We correlated the degree of damage from sputtered atoms to the cohesive energies of the sputtered material. Through fabrication and testing of field-effect transistors, we demonstrate that electron-beam evaporated Sn/Au contacts exhibit superior performance including reduced contact resistance (~12×), enhanced mobility (~4.3×), and lower subthreshold slope (~0.6×) compared to their sputtered counterparts. Our findings underscore the importance of contact fabrication methods for optimizing the performance of 2D MoS2 devices and the value of Raman spectroscopy with backside illumination for gaining insight into contact performance. [ABSTRACT FROM AUTHOR]

Two-dimensional (2D) molybdenum disulfide (MoS2) holds immense promise for next-generation electronic applications. However, the role of contact deposition at the metal/semiconductor interface remains a critical factor influencing device performance. This study investigates the impact of different metal deposition techniques, specifically electron-beam evaporation and sputtering, for depositing Cu, Pd, Bi, Sn, Pt, and In. Utilizing Raman spectroscopy with backside illumination, we observe changes at the buried metal/1L MoS2 interface after metal deposition. Sputter deposition causes more damage to monolayer MoS2 than electron-beam evaporation, as indicated by partial or complete disappearance of first-order E′(Γ)α and A′1(Γ)α Raman modes post-deposition. We correlated the degree of damage from sputtered atoms to the cohesive energies of the sputtered material. Through fabrication and testing of field-effect transistors, we demonstrate that electron-beam evaporated Sn/Au contacts exhibit superior performance including reduced contact resistance (~12×), enhanced mobility (~4.3×), and lower subthreshold slope (~0.6×) compared to their sputtered counterparts. Our findings underscore the importance of contact fabrication methods for optimizing the performance of 2D MoS2 devices and the value of Raman spectroscopy with backside illumination for gaining insight into contact performance. [ABSTRACT FROM AUTHOR]

In recent years, there has been an increasing shift towards using environmentally friendly renewable resources in marine vessels, replacing traditional diesel generators. However, one of the main challenges faced in renewable energy-driven marine vessels is dynamic load management. The feasibility of a renewable-powered electric marine vessel largely depends on the optimal utilization of renewable resources, and storage is an essential component of the marine electric vessel. This paper proposes a two-stage power management system (PMS) for an electric ferry powered by the fuel cell and battery energy storage systems (BESS). The primary objective of the proposed PMS is to ensure a balance between the generated power and the ferry load by minimizing the consumption of hydrogen (H2) fuel. The first stage of the PMS employs particle swarm optimization (PSO), bacterial foraging optimization (BFO), and a hybrid PSO-BFO algorithm to optimize the fuel cell and battery capacity. This is done so that the generated power can follow the load demand. The second stage of the PMS utilizes the Mamdani rule-based fuzzy logic system (FLS) to match the load demand with the generated power. The hybrid PSO-BFO algorithm optimizes the fuzzy control parameters to meet the dynamic load by ensuring optimal H2fuel consumption and battery state of charge (SOC). To obtain optimal values, the load profile of a conventional ferry is used for the proposed PMS. Based on the optimization results, the optimal capacities are found to be 318 kWh and 317.64 kWh for the fuel cell and BESS, respectively, which are obtained using the hybrid PSO-BFO algorithm. The optimal value of H2fuel consumption during cruising is found to be 18 kg. A simulated model-based approach validates the operation of the proposed PMS. The proposed PMS ensures optimal H2fuel consumption and battery SOC while meeting the dynamic load demands of the ferry. The results obtained demonstrate the effectiveness of the proposed PMS in optimizing the renewable energy-driven marine vessel power system.

Abstract Bangladesh is one of the world’s most susceptible countries to climate change. Global warming has significantly increased surface temperatures worldwide, including in Bangladesh. According to meteorological observations, the average temperature of the world has risen approximately 1.2 °C to 1.3 °C over the last century. Researchers and decision-makers have recently paid attention into the climate change studies. Climate models are used extensively throughout the nation in studies on global climate change to determine future estimates and uncertainties. This paper outlines a perceptible stacking ensemble learning model to estimate the temperature of a tropical region—Cox’s Bazar, Bangladesh. The next day’s temperature, maximum temperature, and minimum temperature are estimated based on the daily weather database collected from the weather station of Cox’s Bazar for a period of 20 years between 2001 and 2021. Five machine learning (ML) models, namely linear regression (LR), ridge, support vector regression (SVR), random forest (RF), and light gradient boosting machine (LGBM) are selected out of twelve ML models and combined to integrate the outputs of each model to attain the desired predictive performance. Different statistical schemes based on time-lag values play a significant role in the feature engineering stage. Evaluation metrics like mean absolute error (MAE), mean squared error (MSE), mean absolute percentage error (MAPE), and coefficient of determination (R2) are determined to compare the predictive performance of the models. The findings imply that the stacking approach presented in this paper prevails over the standalone models. Specifically, the study reached the highest attainable R2 values (0.925, 0.736, and 0.965) for forecasting temperature, maximum temperature, and minimum temperature. The statistical test and trend analysis provide additional evidence of the excellent performance of the suggested model.