Rising electricity costs, coupled with the unreliability of the supply, are big challenges facing higher learning institutions in developing nations, where the use of fossil fuels in centralized power generation systems limits the efficiency of the systems in providing sustainable power services. In the case of Ghana, the challenges are more pronounced in urban-based institutions where the availability of electricity is essential in the conduct of learning, research, and administrative activities. This study outlines the design and techno-economic optimization of the microgrid system, powered by renewable energy, for Accra Technical University (ATU) in Ghana, with the aim of providing reliable energy services while minimizing the cost of energy production and its impact on the environment. Using HOMER PRO, the study developed the microgrid model, where the optimized microgrid is composed of a solar–battery microgrid, comprising 3400 solar panels, Schneider Core inverters, and lithium iron phosphate battery systems, where the simulation results indicate the ability of the microgrid system to generate energy that meets 100% of the energy demand, with the expected energy generated per annum being 3.35 GWh, while the autonomy of the battery is expected to be 68.5 h. Economically, the optimized design realizes a Net Present Cost of GHS 17.1 million and a Levelized Cost of Electricity of GHS 0.350/kWh, which reflects a significant reduction in electricity cost compared to traditional grid-connected supply. The optimized configuration supplies the full campus electricity demand primarily from solar-based generation supported by battery storage, thereby reducing dependence on conventional grid electricity and improving the environmental performance of the institution. Apart from the cost savings, the suggested microgrid design enhances the reliability of electricity supply, reduces the risk of power outages, and realizes a significant reduction in greenhouse gas emissions, which aligns with the renewable energy goals of Ghana. The results prove that renewable energy-based microgrids for university campuses are technically viable and economically justifiable. This paper also presents a model that can be followed for developing reliable, sustainable, and cost-effective microgrids for universities in developing countries.

Abstract The increasing frequency of natural hazards, intensified by climate change, poses substantial challenges to sustainable development worldwide. Northern Pakistan, particularly the Hunza district, is highly susceptible to multiple hazards, including landslides, earthquakes, glacier-induced floods, debris flows, and Glacier Lake Outburst Floods (GLOFs), driven by both climatic and tectonic factors. A multi-hazard assessment is essential to understand the complex interactions between these hazards, offering a comprehensive perspective on risk and facilitating more effective disaster preparedness and mitigation strategies. This study addresses the existing gap in multi-hazard assessments, which are often confined to single-hazard evaluations, by developing an integrated multi-hazard susceptibility map for the Hunza district in Northern Pakistan. The region’s complex topography, active tectonics, and accelerated glacier melting contribute to its high vulnerability to cascading and co-occurring hazards. The integrated assessment utilizes diverse data sources, including topographic attributes, geological, hydro-meteorological, environmental variables, and literature-derived hazard map for multi-hazard susceptibility analysis. A Machine Learning (ML) Forest-Based Classification and Regression (FBCR) model, Analytical Hierarchy Process (AHP), and Vs30-based site characterization was employed to classify and generate hazards individually and as integrated multi-hazard susceptibility map. The model incorporates eighteen geo-environmental variables for individual hazards assessment. The resulting multi-hazard susceptibility map indicates that 23.11% of the area is prone to landslides, 6.07% to flash floods, 4.66% to debris flows and flash floods, and 3.98% to a combination of flash floods, landslides, and debris flows. The highest multi-hazard zone, comprising seismic hazard, debris flows, landslides, and flash floods, covers 2.88% of the area, whereas low-hazard zones constitute 56.84% of the region. The landslide susceptibility model classifies 20% of the area as very high susceptible, while the flash flood, debris flow, and seismic hazard models indicate 5, 2, and 13% of the area, respectively, fall under very high susceptibility/hazard. This integrated multi-hazard approach provides a comprehensive risk assessment framework, supporting evidence-based disaster risk reduction policies and infrastructure planning in hazard-prone regions. The findings identify critical high-hazard zones, offering data-driven insights for targeted mitigation strategies and disaster risk reduction efforts.

This study presents an in-depth examination of machine learning (ML) methods to mitigate security risks in the Internet of Medical Things (IoMT). As a result of the faster adoption of innovative wireless technologies in healthcare to assist with remote surgeries, real-time monitoring, and AI-based diagnostic solutions, IoMT systems are more vulnerable to possible cyber threats that can compromise patient information and safety in general. This paper compares the performance of various ML algorithms in countering four major attack vectors, namely Man-in-the-Middle (MITM), DDoS, zero-day attacks, and adversarial attacks. The paper finds trade-offs between detection accuracy, computational complexity, and resistance to adversarial manipulation. Random Forest (RF) and Convolutional Neural Networks (CNN) have higher detection rates but cannot be implemented on resource-limited medical devices. Lightweight algorithms offer efficiency in terms of operations at the cost of accuracy. We identify weaknesses in primary research and propose future research directions, including edge-deployable ML deployments, hybrid solutions with complementary algorithmic advantages, and standardized testing environments of IoMT security solutions. This initiative offers critical information to improve the security posture of next-generation medical networks to protect sensitive health data and maintain the integrity of life-sensitive healthcare services.

The increasing cybersecurity risks confronting Industrial Internet of Things (IIoT) networks present significant challenges to conventional Intrusion Detection Systems (IDS), particularly in terms of computational demands, adaptability constraints, and interpretability. This paper introduces the Adaptive Hybrid Anomaly Detection with Explainable Features (AHADE) model, an innovative intrusion detection framework designed to overcome the challenges of identifying both established attack signatures and previously unknown threats in IIoT environments with limited resources. AHADE achieves this by integrating unsupervised variational autoencoder-based anomaly detection with supervised deep learning classification, capitalizing on the strengths of both approaches. The AHADE architecture utilizes a streamlined variational autoencoder for anomaly detection based on reconstruction, enabling the identification of emergent threats. Complementing this, a lightweight supervised classifier is employed for the accurate recognition of established attack signatures. These components are integrated via an adaptive fusion mechanism, which dynamically adjusts the weighting of predictions based on confidence scores and the characteristics of the detected threats. Extensive evaluation using the Edge-IIoTset, SWaT, and X-IIoTID datasets reveals that AHADE attains robust competitive performance, exhibiting 92.40

Machine learning models, especially vision transformers in the domain of medical images, are highly prone to data poisoning attacks, in which a small proportion of adversarial samples is injected into the model's training dataset to manipulate its behavior. Existing data poisoning techniques have their limitations in terms of the presence of noticeable artifacts in the injected samples or their vulnerability to preprocessing transformations. Similarly, most defence techniques have their limitations in terms of robustness to different types of poisoning attacks. To overcome these challenges, this research presents a structurally embedded invisible poisoning attack technique and a Holistic Defence Strategy (HDS) for the Pyramid Vision Transformer (PVT) model. In this research, the proposed invisible poisoning attack technique takes advantage of the structural characteristics of images, specifically the edges of images, as invisible carriers of trigger information. The proposed attack technique develops a Deep Multi-Scale U-Net Injection Network (DMS-UNet-IN) to embed the trigger information in images in an invisible manner. The proposed attack technique differs significantly from traditional trigger-based techniques in terms of the alignment of perturbations with structural manifolds. The Holistic Defence Strategy (HDS) develops a discriminative detection boundary in the feature space using a mimic model with an attention-aware generative adversarial network. The proposed invisible poisoning attack technique was evaluated using Radiology, Ophthalmology, and Pathology datasets. The experimental evaluation of the proposed invisible poisoning attack technique showed its superiority in terms of invisibility, with a PSNR value of 43.46 dB and SSIM value of 0.9925, over the state-of-the-art techniques such as DeepPoison (40.91 dB) and SPM (38.16 dB). The proposed Holistic Defence Strategy was evaluated using static poisoning rates ranging from 5 to 30%. The experimental evaluation of the proposed Holistic Defence Strategy showed its superiority over the state-of-the-art techniques such as CD, DUTI, and TRIM in terms of detection accuracy up to 95.8% with an F1-score of more than 0.95. The superiority of the proposed technique was also evident in terms of up to 17% improvement over the baseline techniques at higher levels of contamination.

The Internet of Things (IoT) and machine learning (ML) have various applications in different sectors of life, such as healthcare, agriculture, industries, transportation, smart cities, smart homes, etc., and their number is increasing with each passing day. The rapid development of IoT and its increasing demand in different fields of life create a serious problem of security for the IoT environment, which needs serious consideration to protect the IoT-enabled systems from external networks and cyber-attacks. Because of the open deployment environment and constrained resources, the IoT is prone to malicious assaults. Furthermore, the IoT’s diverse and dispersed properties make it difficult for conventional intrusion detection systems (IDS) to keep up with current technological developments. An ML-enabled IoT-based IDS is one of the most important security methods that can assist in defending computer networks and the IoT environment from numerous attacks and malicious activities. Keeping in mind the significant contribution of ML to securing the IoT environment, we proposed an ML-enabled IDS for securing the IoT networks and applications in this study. In the proposed system, we proposed a modified Random Forest (RF) algorithm and compared its performance with nine well-known ML algorithms for the detection of network attacks. Further, two of the most recent and well-known network datasets, i.e., TON-IoT and UNSW-NB15, are used to check the effectiveness of the ML-enabled IDS. The performance of the utilized ML algorithms was measured with the help of different performance measures such as accuracy, sensitivity, etc. The experimental outcomes illustrate the importance of the proposed ML-enabled IDS for securing the IoT environment and applications. The proposed system applies to almost all of the resource-constrained devices that use the IoT network.

The Internet of Underwater Things (IoUT) is a next-level idea for Underwater Acoustic Sensor Networks (UASNs) that are made up of linked nodes that can function and communicate in not only complex but also hard-to-predict aquatic environments. Yet, the deployment of IoUT systems is greatly limited by a number of issues that have not been solved. Such as, the shortage of vital power sources, constantly changing and bandwidth-limited acoustic channels, long transmission delays, high rates of link unreliability, and inaccurate node localization. Current solutions for node localization and data routing often fail to sufficiently tackle these limitations, which leads to a higher rate of redundant transmissions, significant energy wastage, and decreased packet delivery efficiency in fluctuating underwater operational scenarios. This article presents Localization and Energy Efficiency Optimization (LEEO) system, a comprehensive 6G and quantum-inspired solution aimed at the identified problems. The quantum-aware solver in this system improves 6G orchestration by changing the classical algorithms to account for quantum-inspired parameters. At the same time, the Nearest Node Verification (NNV) mechanism verifies that the cooperative nodes selected for data forwarding are not only geographically optimal but also energy-efficient. LEEO also features the following three cooperation-enhancement mechanisms: (i) energy-aware relay rotation with lightweight amplification, (ii) adaptive duty-cycling with wake-up signaling, and (iii) opportunistic short-range optical bursts for localized exchanges to raise channel efficiency without the burden of significant computational overhead on underwater nodes. The 6G surface/Non-Terrestrial Network (NTN) layer enables the integrated sensing and orchestration. Thus, NNV thresholds and relay schedules optimize with the help of topological cues, and quantum-inspired computational techniques used for fast global decisions about routing, relay selection, and rate/power scheduling. The LEEO system significantly reduces the energy consumption per bit, increases the packet delivery ratio, reduces the routing overhead, lowers the errors in node localization, and shortens the end-to-end communication delay as compared to traditional IoUT systems, according to the simulation results. As a result, these findings make LEEO a major step forward in the direction of environmentally friendly, location-aware, and energy-efficient underwater telemetry.

Short-term photovoltaic (PV) power forecasts are essential for storage dispatch, reserve scheduling, and grid safety, yet remain challenging under rapid irradiance ramps and seasonal regime shifts. We present a compact, causal CNN–LSTM architecture that couples local temporal pattern extraction with long-range sequence memory, augmented by physics-aware features (solar geometry, plane-of-array irradiance, clear-sky indices) and strict leakage safeguards. Using a one-hour-ahead task, we evaluate on a 2023 Accra, Ghana simulation study built with PVWatts v8 driven by NSRDB PSM v3.2 (60 kWp DC, 55 kW AC). Metrics are reported in kW and normalized to DC capacity, with daylight/overall splits for fairness. The proposed model achieves RMSE = 0.127 kW, MAE = 0.092 kW, and R^2 = 0.956 on the test split, reducing RMSE by 21.6 k=5 , m=32 ) with d=128 LSTM units is near-Pareto-optimal (about 0.093 M parameters and 2.20 M MACs per step). Baselines (persistence, clear-sky-scaled smart persistence, and GBRT) are included to contextualize deterministic accuracy and skill. We also provide error anatomy by hour and season to highlight residual risks at dawn/dusk and during fast cloud transients. While results are strong, they reflect a simulation (plain PVWatts; no row-to-row shading or sensor noise). We outline a path to operational validation on measured plant AC data across seasons/sites and discuss extensions to probabilistic forecasting with calibrated intervals.

Remote communities often lack access to reliable electricity. This study investigates the feasibility of a microgrid system tailored for Kantong Kunda, a rural community in The Gambia. The community's current energy consumption and demand are determined through data collection using the Epicollect5 survey tool to characterize the local energy consumption and demand profile accurately. HOMER Pro software was employed to simulate and optimize hybrid microgrid configuration, prioritizing both cost-effectiveness and environmental sustain-ability. The proposed system integrates 79.8 kW of Solar Photovoltaic (SPV), a 60-kW diesel generator, 374 batteries, and a 22.8 kW converter. The optimized design yields a Net Present Cost (NPC) of $251,474.80 and a Levelized Cost of Energy (LCOE) of $0.08527/kWh, which is well below the region's grid electricity tariff. This configuration yields 16.1 % excess electricity, a 10.5 % Return on Investment (ROI), a 14.2 % Internal Rate of Return (IRR), and a 6.03-year payback period, while cutting total emissions by 133,981.4 kg compared to a diesel-only baseline. The work contributes a microgrid design designed for rural African communities, and the findings demonstrate that microgrids can deliver reliable, affordable, and low-carbon electricity through decentralized energy systems for remote communities.

Breast cancer remains a leading cause of mortality among women in low- and middle-income countries (LMICs), compounded by fewer radiologists available and resources for diagnosis. A narrative review that summarizes 45 peer-reviewed publications to date from 2018 to 2025 is presented for deep learning (DL) models for mammography detection for breast cancer, targeting low-resource-critical architectures for development in LMIC settings. We compare convolutional neural networks (CNNs), hybrid CNN-support vector machine (SVM) models, recurrent/LSTM networks, and lightweight architectures such as MobileNet and EfficientNet. In quantitative synthesis the reported diagnostic accuracy for MobileNet is between 89 and 92

Low visibility conditions, particularly those caused by fog, significantly affect road safety and reduce drivers' ability to see ahead clearly. The conventional approaches used to address this problem primarily rely on instrument-based and fixed-threshold-based theoretical frameworks, which face challenges in adaptability and demonstrate lower performance under varying environmental conditions. To overcome these challenges, we propose a real-time visibility estimation model that leverages roadside CCTV cameras to monitor and identify visibility levels under different weather conditions. The proposed method begins by identifying specific regions of interest (ROI) in the CCTV images and focuses on extracting specific features such as the number of lines and contours detected within these regions. These features are then provided as an input to the proposed hierarchical clustering model, which classifies them into different visibility levels without the need for predefined rules and threshold values. In the proposed approach, we used two different distance similarity metrics, namely dynamic time warping (DTW) and Euclidean distance, alongside the proposed hierarchical clustering model and noted its performance in terms of numerous evaluation measures. The proposed model achieved an average accuracy of 97.81%, precision of 91.31%, recall of 91.25%, and F1-score of 91.27% using the DTW distance metric. We also conducted experiments for other deep learning (DL)-based models used in the literature and compared their performances with the proposed model. The experimental results demonstrate that the proposed model is more adaptable and consistent compared to the methods used in the literature. The proposed method provides drivers real-time and accurate visibility information and enhances road safety during low visibility conditions.

Because of polypharmacy and complicated treatment protocols, adverse drug reactions (ADRs) continue to be a major problem in oncology and frequently lead to serious clinical complications. Recent developments in the use of artificial intelligence (AI) and machine learning (ML) for ADR prediction in anticancer therapy are critically assessed in this review. We go over a variety of methods for utilizing both structured and unstructured clinical data, such as supervised, unsupervised, and deep learning models in addition to natural language processing (NLP) strategies. Strong performance has been demonstrated by ensemble techniques like Random Forest and Gradient Boosting, while deep neural networks allow for sophisticated feature extraction, albeit with interpretability issues. We highlight new integrative techniques based on current literature trends, such as integrating demographic information, treatment history, and physiological signals with CNN-based models and SHAP-based.

This study investigated the feasibility and sustainability of standalone hybrid energy systems for rural electrification in Ghana. The problem addressed was the lack of electricity access in rural areas of Ghana, despite progress in increasing access rates in urban areas. The importance of this study lies in identifying a reliable, affordable, and environmentally sustainable solution to bridge the electrification gap. This study employed a comprehensive analysis of different system configurations, including Solar Photovoltaic (SPV), Diesel Generators (DG), and Battery Storage (BS), using the Hybrid Optimization of Multiple Electric Resources (HOMER) software. The integration of machine learning models into the design process was also explored to enhance the accuracy of energy consumption predictions. The economic, technical, and environmental aspects of each alternative are evaluated. The significant results obtained from the analysis showed that the optimal design for the standalone hybrid energy system consisted of a 112 kW SPV array, 80 kW DG, 342 kWh BS, and a 54-kW converter. The Levelized Cost of Energy (LCOE) obtained from this configuration was $0.0263/kWh, with a renewable fraction of 91.5%. The Net Present Cost (NPC) of the system was also the lowest among the alternatives, amounting to $244,937. Further investigation of the Electric Distance Limit (EDL) of the optimal solution is 7.5 km which depicts that, although grid-connected electricity may seem more affordable, the proposed standalone hybrid energy system is a more viable option for remote communities that are far from the grid and makes it a suitable solution for addressing the electricity access gap in Ghana's rural areas.

Fault ruptures induced by earthquakes pose a significant threat to constructions, particularly underground structures such as pile foundations. Among various foundation types, batter pile foundations are widely used due to their ability to resist inclined forces. To gain new insights into the response of batter pile groups to fault ruptures caused by earthquakes, this study investigates the deformation and failure mechanisms of batter pile groups due to the propagation of normal and reverse fault ruptures using 3D numerical modeling. An advanced hypoplastic constitutive model for clay, which accounts for small-strain stiffness, and a concrete damage plasticity (CDP) model are employed to simulate the soil and the batter pile foundation, respectively. Results show that following fault propagation, nearly 10% tilting and significant displacement occurred at the pile cap, indicating a total failure of the batter pile foundation. It was also observed that the piles bent towards the slipping direction of the hanging wall. Tensile damage to the pile foundation was notably more severe than compression damage. The most severely damaged regions were not only located at the joints between the piles and the pile caps but were also found along the pile shafts.

Medical robotics is a field that combines engineering and artificial intelligence to create robotic systems used in healthcare. This study aims to delve into the exciting field of developing medical-targeted drug delivery systems, which hold immense potential for revolutionizing how we administer therapeutic substances. By precisely delivering drugs to specific sites within the body, these innovative systems offer improved treatment outcomes, reduced side effects, and enhanced patient compliance. Therefore, an image processing algorithm for visible target detection was designed, a system to determine the coordinates of the exact location of identified regions of the target was developed, and a trajectory plan for the robotic arm end effector for precise drug delivery was also designed. The integration of robotic systems into MRI scanners, the development of image processing algorithms, and also real-time feedback control systems are mechatronics-based processes. Medical image navigation technology has many uses in minimally invasive robots, mainly image acquisition techniques such as Magnetic Resonance Imaging (MRI) to acquire equivalent image data; however, medical imaging techniques have inefficient visibility in robotic-assisted surgeries of small and confined incisions. Therefore, image processing methods are applied to precisely regulate the operation’s location and visibility and improve the procedure’s efficacy, accuracy, and safety. To develop an advanced medical-targeted drug delivery system.

To enhance the fuzzy inference capability of Stochastic Configuration Networks (SCNs), we propose a new neuro-fuzzy model based on Fuzzy Stochastic Configuration Networks (F-SCNs). Unlike traditional SCNs, F-SCNs replace hidden layers with Takagi-Sugeno (T-S) fuzzy inference modules, enabling them to process fuzzy input data, generate meaningful fuzzy rules connected to the output layer and perform reasoning more effectively. A key challenge is establishing a framework that ensures robust modeling performance. To address this, we introduce Self-Organizing Fuzzy Stochastic Configuration Networks (SO-FSCNs) with a Hybrid Learning Algorithm (HL-SOFSCN) for nonlinear system modeling. Additionally, we propose a growing-and-pruning productive approach that refines fuzzy rules based on network knowledge and rule firing intensity. The learning performance of fuzzy rules is improved using error correction algorithms with appropriate initial parameters, while redundant rules with low firing strength are eliminated to maintain a compact structure. Furthermore, we develop a hybrid learning algorithm that integrates a least squares method for parameter tuning with an enhanced second-order optimization approach, treating linear and nonlinear parameters separately to improve learning efficiency. The model is validated using artificial datasets, demonstrating that SCNs achieve a satisfactory predictive accuracy compared to alternative models.

Polycystic Ovarian Disease (PCOD) is among the most prevalent endocrine disorders complicating the health of innumerable women worldwide due to lack of diagnosis and appropriate management. The diagnosis of PCOD, along with proper classification with the help of ultrasound imaging, would be of immense importance for early intervention and timely management of the condition. However, most of the existing approaches suffer from lots of problems, including low accuracy and capability in feature extraction, and may also be resilient to noise; it can further delay or lead to a wrong diagnosis. The main objective of this paper is to address these important issues by proposing a deep learning model, Holographic Convolutional Dense Network (Coco-HoloNet) that will be tailored for the precise detection and classification of PCOD in ultrasound images with high accuracy. These are multi-fold contributions which focus on improvement in diagnostic accuracy by overcoming the various limitations of conventional approaches. CoCo-HoloNet is using a layered architecture by integrating convolutional layers, dense blocks, and pooling strategies that leverage capturing and extraction of significant features from the input effectively. More importantly, the model is also embedded with the Tangent-Runner Adaptive Optimization (TRAdO) technique, which dynamically calculates the regularization parameters to overcome overfitting problems and improves the generalization capability of the model. The approach not only ensures the richest possible feature representation, but it also results in outstanding improvements within the performance measures of a model, such that the accuracy rate exceeds 99%. Further experimentation with CoCo-HoloNet on an extended Kaggle PCOD ultrasound image dataset proves its effectiveness by reporting higher precision, recall, and F1-scores than those obtained by state-of-the-art existing methods.

Wireless Internet of Things (IoT) Sensor Networks (WIoTSNs) are frequently deployed in resource-constrained environments where security threats pose significant challenges. Existing intrusion detection systems (c) often struggle with scalability and efficiency under the unique demands of IoT networks. This work introduces an Intrusion Detection System (IDS) framework that integrates Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks in a hybrid architecture, enhanced by an attention mechanism to improve feature extraction and classification accuracy. To address computational demands, an enhanced Particle Swarm Optimization (PSO) algorithm is implemented for dynamic feature selection, thereby optimizing the system’s efficiency in high-dimensional data environments characteristic of IoT networks. The proposed model enhances IoT intrusion detection by integrating a novel hybrid CNN-LSTM with an attention mechanism, thereby improving feature extraction and temporal pattern recognition. Additionally, the improved dynamic PSO algorithm optimizes feature selection in real time, enhancing classification accuracy and adaptability to evolving IoT network threats. This combination ensures more efficient and robust intrusion detection in dynamic IoT environments. Experimental evaluations using a standard IoT intrusion dataset indicate that the proposed model achieves notable accuracy rates of 98.73% with CNN, 99.87% with LSTM, 99.12% with CNN-LSTM, and 98.88% with the enhanced CNN-LSTM with attention, demonstrating an improvement over existing techniques. The framework’s resilience and adaptability underscore its potential for enhancing network security in real-world IoT applications by addressing evolving threats and computational constraints.

This research work focuses on conceptualizing and modeling smart grid infrastructure. The project's main objective is to develop a smart grid infrastructure to improve energy management in industrial processes. Regenerate has deployed a system that uses renewable energy sources like solar photovoltaic (PV) arrays and wind turbines combined with batteries for energy storage. This research work describes a method that first creates synthetic data and then uses a polynomial function to precisely model the statistical fluctuations inherent in the output of renewable energy sources. The process is carried out using MATLAB. A simulation model created using Simulink is also applied to build a smart grid that includes dynamic load control, demand side management DSM, and various control system configurations. The simulation demonstrates that it is possible to produce electricity with minimal or no depreciation related to an increased dependency on renewable energy sources. It can also function at a higher ratio of energy output produced by traditional sources than previously credited within New Zealand while having substantially fewer overall expenses associated with carbon emissions.