Access to reliable electricity is essential for delivering quality healthcare. However, off-grid health facilities in rural regions like Kalangala, Uganda, often face persistent power outages and high operational costs due to dependence on diesel generators. This study proposes a fuzzy logic-based energy management system (FLC-EMS) to optimize power flow in a hybrid renewable energy system (HRES) combining solar photovoltaics (PV), wind turbines (WT), and battery storage. The system was modeled in MATLAB/Simulink, using 27 fuzzy IF-THEN rules and triangular membership functions to manage four switching ports that prioritize renewable energy based on real-time load demand, renewable availability, and battery state-of-charge (SOC). Simulation results showed that the FLC-EMS ensured continuous power supply during peak demand periods (e.g., 9:00 AM and 7:00 PM) by dynamically balancing solar, wind, and battery inputs. The optimized PV-WT-BAT configuration achieved a Levelized Cost of Electricity (LCOE) of $0.281 and a Net Present Cost (NPC) of $269,246 over a 20-year period. Compared to diesel-based systems, it reduced operational costs by 11.87-18.7% and significantly lowered carbon emissions. The proposed FLC-EMS demonstrates strong potential to improve energy reliability and cost-effectiveness for off-grid healthcare facilities. Its adaptability to variable loads and intermittent renewable sources makes it a scalable solution for sustainable rural electrification. Future research will focus on real-world implementation and enhancing predictive control through machine learning.
This research addresses the challenges of high Peak-to-Average Power Ratio (PAPR), sideband leakage, and spectrum efficiency in 5G wireless networks. We compare Universal Filtered Multi-Carrier (UFMC), Scrambled UFMC (S-UFMC), and Orthogonal Frequency Division Multiplexing (OFDM) techniques using Amplitude Phase Shift Keying (APSK) modulation. Our findings show that APSK modulation significantly reduces PAPR compared to traditional methods. Integrating Particle Swarm Optimization (PSO) with Partial Transmit Sequences (PTS)-OFDM and S-UFMC minimizes PAPR and computational complexity. Results demonstrate that S-UFMC with PSO optimization achieves superior performance, offering lower PAPR, reduced complexity, and enhanced spectral efficiency, positioning it as a promising 5G waveform. This research highlights the potential of advanced waveform designs to improve 5G communication systems. UFMC demonstrated improved spectral efficiency with sidelobe levels reaching as low as 0.2, indicating efficient spectrum utilization. The research achieved a PAPR of 7.393 dB for 16-APSK with 512 subcarriers and 7.414 dB for 1024 subcarriers after optimization. The PSO algorithm significantly reduced PAPR values, with the Scrambled UFMC system outperforming PTS-OFDM and standard UFMC in terms of PAPR.
Industrial waste-based catalysts provide a sustainable and cost-efficient solution for biodiesel production, improving yield, quality, and environmental impact. When this biodiesel is used in advanced reactivity-controlled compression ignition (RCCI) mode, it enhances the combustion process within direct injection (DI) diesel engines. These strategies effectively reduce nitrogen oxide (NOx) emissions and smoke without compromising engine performance. This study used cottonseed (Gossypium arboreum) methyl ester (CSME) as the pilot injection fuel. It was produced under optimal conditions of 2 wt% industrial waste dolomite catalyst, an 8:1 methanol-to-oil molar ratio, and heating at 55 °C for 45 min during transesterification through the response surface methodology (RSM) with central composite design (CCD). The catalytic potential of the industrial waste dolomite catalyst is validated through X-ray diffraction (XRD), Fourier-transform infrared spectroscopy (FTIR), and Brunauer-Emmett-Teller (BET) analyses. Next, the n-butanol was injected into the intake manifold of the diesel engine at different energy shares of 10%, 20%, and 30% using an electronic primary fuel injection (EPFI) system in the RCCI mode. The fuel blends of diesel, CSME10 (10% CSME + 90% diesel), CSME20 (20% CSME + 80% diesel), and CSME100 (100% CSME) were tested as single-fuel in conventional mode, and CSME100 + 10% n-butanol, CSME100 + 20% n-butanol, and CSME100 + 30% n-butanol were tested in RCCI mode under variable load settings. Compared to the single-fuel operation, the RCCI combustion mode improved the performance and reduced emissions characteristics for all n-butanol energy shares. Especially, the CSME100 + 30% n-butanol mixture boosts brake thermal efficiency (BTE) by 22.25% and lowers brake specific fuel consumption (BSFC) by 23.33%. The unburnt hydrocarbon (HC) and carbon monoxide (CO) emissions were slightly increased by 28.13% and 27.37%, respectively. Also, the RCCI mode could simultaneously reduce smoke opacity (up to 58.07%) and NOx emission (up to 41%) through lower peak cylinder pressure and heat release rate (HRR) at 18 kg in 100% engine load operation. Based on these analyses, it is suggested that the RCCI mode with n-butanol injection by the EPFI system shows efficient fuel combustion and significantly reduced tailpipe emissions in DI diesel engine applications.
26 September 2025 in BMC Medical Informatics and Decision Making
Colorectal cancer (CRC) continues to be a significant global health issue as a result of the difficulties associated with the early and precise detection of lesions through colonoscopy. We propose CADxPolyDetect, a clinically explainable hybrid deep learning system for multi-class colorectal lesion detection using augmented colonoscopy images, to address diagnostic limitations that arise from class imbalance, interpretability gaps, and spatial complexity in colonoscopic images. A five-stage experimental pipeline is integrated into the framework. Initially, the original Hyper Kvasir dataset (10,672 images) is augmented to 23,000 images using Deep Convolutional Generative Adversarial Networks (DCGAN) and class balanced using SMOTE. Three pre-trained CNNs—ResNet-50, DenseNet-201, and VGG-16—are implemented in Stage 2 to execute end-to-end feature extraction. Transformer networks are integrated with these CNNs in Stage 3 to capture global spatial dependencies. Stage 4 incorporates multi-class support vector machines (SVM) for the final classification of 23 colorectal conditions. Stage 5 improves interpretability by utilising Grad-CAM to produce clinically relevant heatmaps that localise malignant polyps and facilitate transparent decision-making. The DenseNet-201 + Transformer + Multi-Class SVM model showed the best results of all the setups tested, achieving an overall accuracy of 98%, an F1-score of 0.98, a precision of 97%, and a recall of 99%, based on a 70:30 split of training and testing data. Robustness, precision, and clinical trust are guaranteed through the integration of attention-based visualisations, deep ensemble learning, and augmented data. Therefore, CADxPolyDetect sets a new benchmark in AI-supported colonoscopy by offering a decision support system that is easy to understand and can grow to meet needs for the early detection and screening of colorectal cancer.
Deep learning models are frequently implemented as black boxes, which restricts their application in sensitive sectors such as agriculture, despite their ability to achieve high classification accuracies. To resolve this issue, we introduce a comprehensive interpretability framework that implements Grad-CAM, Grad-CAM++, and LIME techniques on the most effective CNN + ViT + SVM ensemble model for the detection of citrus diseases. The proposed model obtained classification accuracies of 99% and 97% by augmenting the real-world Lemon and Orange datasets with 1600 and 4000 images, respectively. Interpretability was implemented on the DenseNet121 (Orange) and InceptionV3 (Lemon) backbones. The model focused on specific textures related to diseases, which was shown by the detailed image analysis from LIME, while Grad-CAM and Grad-CAM ++ highlighted clear areas of lesions with high detail. A perfect score of 1.00 over all assessed courses was achieved through quantified assessment using Intersection over Union (IoU) and Dice Coefficient, and confirmed accuracy in visual explanation. In addition, intra-class compactness and inter-class separability were confirmed using feature-space t-SNE mappings. This visual explanation framework enhances transparency and builds user trust in machine learning-based disease management decision making in practice through clearly interpretable visual explanations for both multi-class and binary classification tasks. This research is a major advance towards developing trustworthy and comprehensible AI (XAI) solutions for agricultural diagnosis, through providing visual explanations to both multi-class and binary classification tasks. The approach is scalable and model-independent and is capable of integration into real-time and mobile decision support tools.
With the rapid growth of multimodal data on the web, cross-modal retrieval has become increasingly important for applications such as multimedia search and content recommendation. It aims to align visual and textual features to retrieve semantically relevant content across modalities. However, this remains a challenging task due to the inherent heterogeneity and semantic gap between image and text representations. This paper proposes mutual contextual relation-guided dynamic graph network that integrates ViT, BERT and GCNN to construct a unified and interpretable multimodal representation space for image-text matching. The ViT and BERT features are structured into a dynamic cross-modal feature graph (DCMFG), where nodes represent image features and text features, and edges are dynamically updated based on mutual contextual relations that is neighboring relations extracted using KNN. The attention-guided mechanism refines graph connections, ensuring adaptive and context-aware alignment between modalities. The mutual contextual relation helps in identifying relevant neighborhood structures among image and text nodes, enabling the graph to capture both local and global associations. Meanwhile, the attention mechanism dynamically weights edges, enhancing the propagation of important cross-modal interactions. This emphasizes meaningful connections or edges among different modality nodes, improving the interpretability by revealing how image regions and text features interact. This approach overcome the limitations of existing models that rely on static feature alignment and insufficient modeling of contextual relationships. Experimental results on benchmark datasets MirFlickr-25K and NUS-WIDE are used to demonstrate significant performance improvements over state-of-the-art methods in precision and recall, validating its effectiveness for cross-modal retrieval.