Browse Articles

Adaptive deep reinforcement learning–based secure routing for wireless sensor networks

Research on loss optimization calculation and cooling structure design of high-speed permanent magnet motor rotor for integrated starter-generator

Phishing webpage detection using structured URL generation

Imbalance fault diagnosis based on iterative attention wavelet packet decomposition and weighted ensemble classification algorithm

Prioritising construction accident risk factors using an integrated Random Forest, XGBoost, and ANN framework with sensitivity analysis

TrafCopAgent: synergizing reinforcement learning and multi-agent collaboration for adaptive emergency traffic control

Diagenetic processes in quaternary vertebrate fossils from coastal environments

Abstract The fossilization of Quaternary mammals in coastal environments of southern Brazil results from the interaction between taphonomic and diagenetic processes, strongly controlled by hydrological dynamics, interstitial fluid chemistry, and sedimentary reworking. Through the integration of macro- and microscopic analyses with SEM, XRD, and Raman spectroscopy, two main diagenetic pathways — Route A and Route B — were identified. Route A develops in silica-rich environments, characterized by mechanical sediment infiltration and precipitation of siliceous phases. These processes promote increased structural disorder in apatite, reduction in crystallite size, pronounced removal of B-type carbonate, and the development of vibrational bands associated with phosphate–silicate interactions. In parallel, Route B is associated with carbonate-supersaturated solutions. In this pathway, calcite precipitation and the formation of larger crystallites indicate a buffered geochemical system capable of delaying apatite decarbonation. Evidence of this buffering is preserved in Raman data, which reveal a diagenetic gradient marked by narrowing of the ν₁(PO₄³⁻) band (decreased FWHM) together with progressive but partial depletion of B-type carbonate. XRD data confirm the formation of fluorapatite through diagenetic substitution of OH⁻ by F⁻, as well as variations in crystallite size consistent with distinct fluid–mineral interaction histories. Taphonomic evidence indicates variable redox conditions and intense pore-fluid circulation. The distinction between Route A and Route B provides a useful interpretative framework for understanding bone preservation in Quaternary coastal environments.

LIR-Mamba: consolidating YOLO and selective SSM with global-local scanning for robust infrared small target detection under laser interference

Hydrodynamic characteristics during floating and installation of high blockage-ratio immersed tunnels in Inland Rivers

A graph-guided cross-modal attention network for multimodal emotion recognition via emotion-shift modeling

Integrated high density seismic imaging and structural characterization of the Yahu deep brine system, Qaidam Basin

Explainable hybrid deep learning for automated cervical cytology classification

Abstract Accurate and explainable automated analysis of cervical cytology remains a major challenge for artificial intelligence (AI)-assisted cervical cancer screening, particularly in low-resource settings where access to expert cytotechnologists is limited. Although deep convolutional neural networks (CNNs) have substantially improved classification accuracy, their limited interpretability hinders clinical trust and routine adoption. Conversely, neuro-fuzzy systems provide transparent decision-making but lack the ability to learn complex cytomorphological representations directly from image data. Here, we present PapsAI XNet , a novel explainable hybrid deep learning framework that integrates attention-guided CNN feature extraction with an Adaptive Neuro-Fuzzy Inference System (ANFIS) to simultaneously achieve high diagnostic accuracy and interpretable decision-making. PapsAI XNet incorporates a Channel Attention Module and feature normalization strategy to enhance morphological feature discrimination and stabilize downstream fuzzy inference. A compact ANFIS rule base initialized through data-driven clustering enables transparent reasoning while avoiding the rule explosion commonly associated with high-dimensional neuro-fuzzy systems. The framework was evaluated on the publicly available Herlev dataset comprising seven cervical cytology classes and benchmarked against ResNet-18 and MobileNetV2 using accuracy, sensitivity, specificity, precision, F1-score, receiver operating characteristic (ROC) analysis, computational complexity, and explainability analyses. PapsAI XNet achieved an overall accuracy of 97.8% , sensitivity of 96.4% , specificity of 98.6% , precision of 97.1% , and F1-score of 96.7% , significantly outperforming both benchmark CNN architectures. The greatest improvements were observed for mild and moderate dysplasia, where gradual morphological transitions frequently challenge conventional deep learning models. Class-wise area under the ROC curve exceeded 0.97 for all abnormal cytological categories. Furthermore, learned Gaussian membership functions, fuzzy-rule surfaces, and rule activation patterns provided clinically meaningful explanations linking deep morphological features to diagnostic outcomes while maintaining a lightweight computational footprint suitable for real-time deployment. These findings demonstrate that integrating attention-guided deep feature learning with interpretable neuro-fuzzy reasoning enables accurate, explainable, and computationally efficient cervical cytology classification. PapsAI XNet provides a practical framework for trustworthy AI-assisted cervical cancer screening and represents an important step toward clinically deployable explainable AI for digital cytopathology, particularly in resource-constrained healthcare systems.

Comparative assessment of geotechnical and microstructural transformations in clayey soil stabilised with wheat straw ash, pottery waste, and lime

Abstract Expansive clayey soils show high plasticity, swelling and low bearing resistance, limiting pavement and foundation use. This study compares wheat straw ash (WSA), pottery waste (PW) and lime for stabilising high-plasticity clay through performance-based optimisation and mechanism interpretation. Raw materials were characterised using particle-size distribution, index properties, XRF and SEM-EDX, and mixes were prepared on a total dry mass basis. Initial dosage ranges of WSA: 5–20%, PW: 5–30% and lime: 1–6% were screened. Optimum contents were selected using a threshold-based multi-criteria approach based on differential free swell (DFS), pH, consistency limits and compaction behaviour, identifying 12.5% WSA, 22% PW and 4% lime. WSA eliminated DFS and reduced plasticity index to 7.6, mainly through active-clay dilution, pore filling and partial matrix densification. PW reduced plasticity index to 7.01 and improved behaviour through angular-particle packing, interlocking and load transfer. Lime reduced plasticity index to 6.43 and produced the strongest response due to high alkalinity, fabric modification and calcium-assisted bonding. Compaction showed increased optimum moisture content and reduced maximum dry density for WSA and lime, while PW maintained better compactability. Unconfined compressive strength and soaked California bearing ratio tests evaluated the optimum mixes. At 28 days, WSA, PW and lime achieved strength values of 768.3, 869.65 and 1245 kPa and CBR values of 4.29, 4.80 and 9.36, respectively. SEM and XRD observations were consistent with distinct stabilisation responses: active-clay dilution and pore filling for WSA, granular skeleton modification for PW, and calcium-assisted physicochemical modification for lime. The findings show that WSA and PW can serve as useful waste-derived alternatives, although lime remained the most effective mechanical stabiliser.

An updated systematic review, meta-analysis, and network meta-analysis of breast cancer incidence after metabolic and bariatric surgery

Explainable deep learning for early sepsis detection from ICU time-series data using XAI techniques

Abstract Sepsis is one of the most deadly illnesses with a high risk of mortality. Consequently, identifying it at the beginning of illness symptoms is crucial and plays a key role in improving patient outcomes. This study presents a customized solution for the early detection of sepsis with an emphasis on the use of interpretability and explainability techniques, utilizing a range of machine learning approaches and interpretable artificial intelligence methods. The database on which this research study is based has many problems; the main ones being large data gaps and class disparities. Employing robust methods, precise categorizations, and rigorous computations, Approximately 12 diverse models were developed and optimized. With ROC-AUC indicators of 0.9566 and 0.9595 and F1 scores of 0.85 and 0.85 respectively, Bidirectional Long Short-Term Memory (BiLSTM) and Temporal Convolutional Network (TCN) models performed better than conventional models in terms of sepsis prediction. These two approaches have shown remarkable progress in detecting clinical patterns while avoiding false negative results—an essential aspect of the medical field. To assess model performance and offer clear insights into model predictions, interpretation-based techniques were employed. This improved clinical confidence and facilitated well-informed decisions in crucial medical diagnoses.

Computational discovery of bioactive flavonoids from Ziziphus mauritiana via integrated ADME prediction, molecular docking, DFT calculation and molecular dynamics simulation

Impacts of anthropogenic and environmental stressors on biotic communities in Al-Mahmoudia Canal, Egypt: a seasonal assessment of water quality and plankton dynamics

Abstract Al-Mahmoudia Canal is crucial for water supply in Egypt’s Al-Beheira and Alexandria governorates, but it faces ecological challenges from pollution. This study examines plankton dynamics and water quality across four seasons at seven sites, analyzing microbiological contamination, heavy metals, turbidity, and nutrients. Phytoplankton abundance peaked in spring at Station 4, with seasonal variations linked to temperature and nutrients. Zooplankton showed seasonal diversity, with increased turbidity and nutrient levels in summer and autumn due to organic pollution. Heavy metal and microbial contamination also fluctuated seasonally, with fecal coliform levels exceeding safety limits. The study highlights pollution hotspots and emphasizes the need for ongoing monitoring and management to ensure sustainable water resource use in the Nile Delta.

Genotype-specific responses of six grapevine cultivars to in vitro propagation

Unified digital biomarker platform for wearable devices assuring integrity, heterogeneity, and scalability

Degradation-aware PSO–MPC energy management for second-life battery prosumer microgrids

Abstract The growing retirement of electric vehicle lithium-ion batteries has created significant opportunities for their reuse in stationary energy storage systems. However, integrating second-life batteries (SLBs) into residential photovoltaic (PV)-based prosumer microgrids remains challenging due to heterogeneous initial state-of-health (SoH) conditions, accelerated degradation behavior, and uncertain remaining useful life. This paper proposes a degradation-aware energy management framework for SLB-integrated prosumer microgrids based on the joint use of particle swarm optimization (PSO) and model predictive control (MPC). The proposed framework consists of four coordinated stages: PSO-based capacity sizing, hybrid AI-based forecasting of PV generation, electricity prices, and load demand, SoH-aware adaptive battery management within the MPC layer, and net present value (NPV)-based battery replacement decision-making. Within the control layer, the adaptive battery management mechanism updates state-of-charge limits, C-rate constraints, and degradation penalty coefficients according to real-time SoH evolution, thereby mitigating excessive capacity fade while preventing economically inefficient underutilization. In addition, a three-stage replacement strategy is developed to determine whether continued operation, SLB, or fresh battery replacement yields the highest long-term economic benefit. The proposed method is evaluated through long-horizon simulations using 2024 meteorological, market, and demand data for five residential prosumers in Antalya, Türkiye. The results show that the framework achieves a fleet-average self-consumption ratio of 71.8%, a renewable energy penetration factor of 37.6%, and a cost of energy of 0.0722 $/kWh, while sustaining battery utilization and enabling lifecycle-aware operation under realistic uncertainty. These findings demonstrate that the proposed PSO–MPC framework provides a computationally tractable approach that achieved stable convergence and feasible MPC solve times (2.1–3.4 s per 72-hour horizon) across all 20 simulated years under the five-prosumer Antalya configuration for degradation-aware energy management of SLB-based microgrids.