River water quality in monsoon-driven subtropical basins exhibits strong seasonal variability driven by hydroclimatic forcing and increasing anthropogenic pressure, posing challenges for reliable assessment and management. Despite advances in water quality modeling, most Water Quality Index (WQI) prediction frameworks require extensive sampling and lack interpretability, limiting rapid baseline assessment during critical periods. This study develops the first integrated Explainable Artificial Intelligence (XAI) framework combining Machine Learning (ML), Deep Learning (DL), and Physics-Informed Neural Networks (PINNs) to predict, interpret, and spatially characterize seasonal water quality dynamics in the Padma River Basin, Bangladesh. Forty-four surface water samples collected during winter and monsoon seasons were evaluated using WQI assessment, explainable modeling, probabilistic uncertainty analysis, and spatial regionalization. Results show that seasonal variability dominates over spatial variability (p < 0.0001), with winter low-flow conditions promoting solute concentration and localized degradation, while monsoon discharge drives basin-wide dilution and recovery. Model performance is strongly region-dependent: Deep Neural Networks achieve the highest accuracy in winter (R2 = 0.98; RMSE = 1.16), whereas Ridge Regression and Voting Ensemble models perform more robustly during the monsoon (R2 ≈ 0.97; RMSE ≈ 1.01). Explainable AI analysis identifies NO3- emerged as the dominant contaminant (24.0 ± 36.3mg/L winter, 47.5 ± 68.7mg/L monsoon, with isolated samples exceeding WHO limits), whereas pH and DO exhibit dual seasonal influences. PINN-based data augmentation improves model generalization under limited sampling while preserving hydrochemical consistency. Monte Carlo simulations quantify prediction uncertainty and reveal seasonal shifts in WQI probability distributions, while spatial autocorrelation analysis identifies localized winter degradation hotspots and widespread monsoon improvement. The proposed physics-informed and explainable AI framework enhances predictive reliability, interpretability, and decision relevance, offering a transferable approach for uncertainty-aware water quality assessment and adaptive management in monsoon-affected, data-limited river basins.

Abstract Minor changes in process variables, such as temperature, processing speed, and cooling rate, can significantly impact the properties of the final product in a sheet extrusion process. As a result, many optimization efforts focus on each of these variables. This study explored a machine learning-assisted process design for a polypropylene/carbon black composite sheet. The extrusion process parameters were selected as input variables, while tensile strength, void content, width, and thickness were measured, resulting in a dataset of 180 entries. A deep learning neural network was employed to identify and propose optimal combinations of process parameters and validate these proposals by comparing predicted values to experimental data. The polymer melt index had a significant effect on tensile strength, which is attributed to the degree of crystallinity. The process optimization led to a 20% increase in the tensile strength of continuous fiber composites, enhancing the matrix's toughness and improving interfacial load-carrying capacity.

With the evolution of Reconfigurable Intelligent Surface (RIS) technology, its potential for dynamically optimizing wireless channels has garnered significant attention. However, existing methods still face challenges in real-time control in complex environments due to high computational complexity. To address this, this paper proposes a reconfigurable wireless channel optimization framework based on Intelligent Metasurfaces 2.0 and designs a low-complexity control strategy. The strategy integrates an adaptive adjustment mechanism and multi-dimensional feedback, aiming to reduce system computational load. Experimental results show that compared to traditional methods (such as MRC and MMSE), the proposed method improves signal transmission quality (SNR improvement of 3.8 dB) and system stability (exponential increase to 0.92). When compared to advanced deep reinforcement learning (DRL) and graph neural network (GNN) methods, it achieves similar signal quality while reducing computational overhead by 20.0% and energy consumption by approximately 32.4%. Ablation experiments further verify the effectiveness and synergistic role of the proposed core modules. This study provides a feasible approach toward high-efficiency, low-complexity dynamic channel optimization in 5G and future communication networks.

Early prediction of lithium-ion battery life using a hybrid deep neural network and ensemble learning approach

OCR: OmniNet-Fusion: A hybrid attention-based CNN-RNN model for multi-omics integration in precision cancer drug response prediction.

Advancing carbon-neutral wastewater treatment: Artificial intelligence-driven strategies for emission mitigation and process optimization.

An effective deep learning and graph neural network approach for accurate prediction of LncRNA-disease associations

Retraction notice to “Hybrid forecasting model based on long short term memory network and deep learning neural network for wind signal” [Appl. Energy 236 (2019) 262–272

Textile and fashion design are among the disciplines where technological advancements are rapidly integrated, with artificial intelligence (AI) playing an increasingly vital role. AI-driven design software enhances fabric texture, color harmony, and pattern creation, while smart textiles and algorithms enable personalized clothing production. Additionally, AI-based analyses are being widely utilized in fashion trend forecasting and sustainable material development. Understanding the evolving research landscape in this field is crucial for future innovations. This study conducts a bibliometric analysis of 856 research articles on textile and fashion design published between 1980 and 2024. Using the Web of Science database and VOSviewer software, the study evaluates literature trends, influential studies, key authors, and emerging keywords. The results indicate that academic interest in this field peaked in 2024, with the most cited study being Dong, K.; Peng, X.; and Wang, ZL’s research on fiber- and fabric-based nanogenerators for wearable electronics and AI applications. China emerges as the leading contributor to AI-driven fashion research, while Chengkuo Lee stands out as a key figure in the field. Keyword analysis highlights strong associations with concepts such as machine learning, artificial intelligence, computer vision, deep learning, and convolutional neural networks. The shift in 2024 towards keywords like “electronic textiles,” “biomarkers,” and “LLM” (large language models) reflects the growing intersection of AI, intelligent materials, and wearable health technologies. These findings underscore the increasing importance of AI in shaping the future of textile and fashion design, necessitating interdisciplinary collaborations. Despite the study’s limitation of relying solely on the Web of Science database and not conducting content analysis, its findings provide a comprehensive overview of AI's evolving role in the sector. This research offers a valuable foundation for future studies, guiding researchers and industry professionals in leveraging AI technologies for innovative design, sustainable production, and enhanced consumer experiences.

Intelligent decision-making for mine ventilation systems based on graph neural network and deep reinforcement learning fusion.

Spiking Neural Networks (SNNs) offer a paradigm of energy-efficient, event-driven computation that is well-suited for processing asynchronous sensory streams. However, training deep SNNs robustly in an online and continual manner remains a formidable challenge. Standard Backpropagation-through-Time (BPTT) suffers from a prohibitive memory bottleneck due to the storage of temporal histories, while local plasticity rules often fail to balance the trade-off between rapid acquisition of new information and the retention of old knowledge (the stability-plasticity dilemma). Motivated by the tripartite synapse in biological systems, where astrocytes regulate synaptic efficacy over slow timescales, we propose Astrocyte-Gated Multi-Timescale Plasticity (AGMP). AGMP is a scalable, online learning framework that augments eligibility traces with a broadcast teaching signal and a novel astrocyte-mediated gating mechanism. This slow astrocytic variable integrates neuronal activity to dynamically modulate plasticity, suppressing updates in stable regimes while enabling adaptation during distribution shifts. We evaluate AGMP on a comprehensive suite of neuromorphic benchmarks, including N-Caltech101, DVS128 Gesture, and Spiking Heidelberg Digits (SHD). Experimental results demonstrate that AGMP achieves accuracy competitive with offline BPTT while maintaining constant O ( 1 ) temporal memory complexity. Furthermore, in rigorous Class-Incremental Continual Learning scenarios (e.g., Split CIFAR-100), AGMP significantly mitigates catastrophic forgetting without requiring replay buffers, outperforming state-of-the-art online learning rules. This work provides a biologically grounded, hardware-friendly path toward autonomous learning agents capable of lifelong adaptation.

Memristors, whose conductance depends on their past electrical history, are the foundation of emerging brain-inspired artificial computing architectures. Here, we demonstrate a unipolar memristor in which both ionic conductance and electroosmotic flow exhibit pronounced hysteresis, enabling dual-mode memory in charge and water transport. Strikingly, this behavior emerges without structural asymmetry or chemical modification. Instead, it originates from a novel mechanism, which involves a reversible transition in a nanoconfined system driven by charge inversion, where counterions overcompensate surface charge. This transition marks a boundary between two distinct electrostatic states in response to an applied electric field. We harness this unique mechanism to emulate synaptic plasticity and implement learning and classification in artificial neural networks and convolutional models. These findings establish a new class of field-tunable aqueous platforms, unlocking opportunities in neuromorphic logic, adaptive computing, biointerfacing, and real-time environmental sensing.

Triply periodic minimal surface (TPMS) structures provide high surface area to volume ratios and tunable conduction pathways, but predicting their thermal behavior across different metallic materials remains challenging because multi-material experimentation is costly and full-scale simulations require extremely fine meshes to resolve the complex geometry. This study develops a physics-informed neural network (PINN) that reconstructs steady-state temperature fields in TPMS Gyroid structures using only two experimentally measured materials, Aluminum and Silver, which were tested under identical heat flux and flow conditions. The model incorporates conductivity ratio physics, Fourier-based thermal scaling, and complete spatial temperature profiles directly into the learning process to maintain physical consistency. Validation using the complete Aluminum and Silver datasets confirms excellent agreement for Aluminum and strong accuracy for Silver despite its larger temperature gradients. Once trained, the PINN can generalize the learned behavior to nine additional metals using only their conductivity ratios, without requiring new experiments or numerical simulations. A detailed heat transfer analysis is also performed for Magnesium, a lightweight material that is increasingly considered for thermal management applications. Since no published TPMS measurements for Magnesium currently exist, the predicted Nusselt numbers obtained from the PINN-generated temperature fields represent the first model-based evaluation of its convective performance. The results demonstrate that the proposed PINN provides an efficient, accurate, and scalable surrogate model for predicting thermal behavior across multiple metallic TPMS structures and supports the design and selection of materials for advanced porous heat technologies.

Abstract This study investigates the use of state-of-the-art software tools available on contemporary desktop computing platforms to enhance predictive modeling with machine learning methods. Existing research has not sufficiently examined how efficient utilization of such tools—specifically state-space search reduction, operation parallelization, and mechanisms for escaping local optima—affects model performance when applied to large-scale high-frequency datasets. To address this gap, we introduce new predictive models that explicitly leverage these advanced software capabilities. We further propose strategies for overcoming local optima in neural-network training and for parameter tuning in population-based metaheuristic algorithms used for forecasting high-frequency financial data. Empirical evaluation is conducted on one-minute EUR/CZK exchange rate data from 2018 and on 17 high-frequency Amazon stock price datasets spanning 2005–2021. The results demonstrate that incorporating modern software optimization tools not only improves predictive accuracy but also significantly reduces computation time, making the approach well-suited for real-time forecasting of highly dynamic financial time series

Artificial intelligence (AI) holds great promise for advancing diagnostics and treatment in nuclear medicine. The rapid growth of AI over the past decade largely driven by advances in hardware components such as graphics processing units (GPUs) and the introduction of Deep Learning (DL) and convolutional neural networks (CNN). The integration of AI and medical imaging has the potential to revolutionize nuclear medicine by, e.g., accelerating image acquisition, enhancing image quality, enabling advanced image generation, assisting image interpretation, and aiding treatment planning. Clinical applications have been demonstrated for most medical specialties, including oncology, neurology and radionuclide therapy. The utilization of AI to provide automated, standardized procedures can help bring advanced imaging from major university centers to smaller local clinics, thus benefiting a broader range of patients. Additionally, AI has vast potential for predicting optimal treatment strategies, assessing risk, optimizing patient flow and outcome, and even improving productivity, but these capabilities have yet to be fully utilized. The fraction of clinical AI applications in general healthcare reaching beyond the prototyping phase are reported as low as 2% [1]. Indeed, in nuclear medicine very few AI developments have reached commercial maturity. Currently, most AI applications in nuclear medicine follow the imaging flow from image acquisition and reconstruction, post-processing and image preparation, image analysis, and decision support for clinical interpretation. Below we will briefly review selected areas and comment on challenges and opportunities for AI in nuclear medicine, with a special focus on the transition from development to clinical implementation.

Classification of Distracted Driving Using Transfer Learning and Deep Neural Network

ABSTRACT Ecological and neural networks share fundamental principles of adaptation and learning. Here, we reformulate the Lotka-Volterra model of species interactions as a neural network model. This transformation enables adaptive modification of interaction coefficients based on the difference between expected (i.e. predicted) and actual population size. This mechanism is analogous to synaptic plasticity in the brain, where neurons minimize the discrepancy between actual and expected activity. Our simulations reveal that ecological systems can ‘learn,’ adjusting interactions to support larger predator populations under changing environmental conditions. This suggests that ecological networks, like neural networks, self-organize through predictive adaptation, which was shown to also maximize metabolic energy. Thus, prediction-based adjustment of interactions may offer a unifying framework for understanding biological complexity across scales: from the single-cell level (i.e. neurons) to the interactions at the level of animal species.

Early identification of students at risk of academic underperformance remains a persistent challenge in higher education, particularly in learning environments characterized by complex, temporally evolving patterns of engagement and assessment. Conventional learning analytics approaches typically rely on static or weakly temporal indicators, limiting their ability to detect emerging risk at early stages of an academic term. This study proposes a sequence-aware learning analytics framework that leverages transformer-based models to represent student academic trajectories as ordered sequences of learning events derived from learning management systems and student information systems. The framework integrates heterogeneous behavioral, temporal, and performance signals and applies self-attention mechanisms to capture long-range dependencies and evolving risk patterns. Using a supervised predictive modeling design with rolling-window and early-prediction evaluation protocols, the proposed approach is assessed against traditional machine learning and recurrent neural network baselines. Results demonstrate that transformer models achieve superior predictive performance, earlier risk identification, and greater stability across academic terms and cohorts. Attention-based interpretability further reveals meaningful progression patterns associated with academic disengagement and performance decline. The findings underscore the value of sequence-aware modeling for enhancing institutional early-alert systems and supporting proactive, personalized academic interventions. This study contributes to both learning analytics theory and practice by establishing transformer-based sequence modeling as a robust foundation for early academic risk detection and student success initiatives in higher education.

Oral cancer is a significant global health concern, particularly in countries like India, where tobacco and betel nut use are prevalent. Despite advances in therapy, the prognosis remains poor, largely due to late-stage diagnosis. Early detection is key to improving survival rates and reducing the burden of treatment. In recent years, Artificial Intelligence (AI) has emerged as a revolutionary tool in medical diagnostics, with promising applications in oral oncology. This article aims to explore the role of AI in the early detection of oral cancer, its current applications, diagnostic accuracy, limitations, and the future direction of its integration into routine oral healthcare. An extensive review of the current literature was conducted, focusing on AI techniques such as machine learning (ML), deep learning (DL), and convolutional neural networks (CNNs), as well as their applications in analyzing intraoral images, radiographs, histopathology slides, and salivary biomarkers. Clinical trials, pilot studies, and technological assessments were reviewed to evaluate the performance of AI in detecting oral potentially malignant disorders (PMDs) and early-stage squamous cell carcinoma. AI-based tools have shown considerable promise in the accurate and non-invasive diagnosis of oral lesions. These systems offer enhanced sensitivity and specificity, reduce human error, and provide objective assessments, even in low-resource or remote settings. DL algorithms, particularly CNNs, have demonstrated excellent performance in image recognition tasks relevant to oral pathology. However, challenges such as data standardization, algorithmic bias, lack of clinical validation, and ethical concerns still hinder widespread adoption. AI has the potential to transform early detection strategies for oral cancer by supporting clinicians in making faster and more accurate diagnoses. With proper validation, integration into clinical workflows, and adherence to ethical guidelines, AI can serve as an invaluable adjunct in oral medicine, especially for mass screening and personalized diagnostics. Continued research, investment in digital infrastructure, and training of dental professionals are essential for realizing its full potential in the future of oral healthcare.

Air pollution poses significant threats to human health and environmental sustainability, requiring strong predictive models to monitor and forecast air quality. This research sought to develop and assess a resilient air pollution forecast model using data-driven modelling methodologies. The study used a thorough technique that included the compilation of worldwide air pollution datasets, succeeded by data pre-treatment and modification to guarantee the precision and pertinence of the input data. This data-centric methodology enabled the examination and interpretation of the dataset using several machine learning methods. The research examined the efficacy of several machine learning algorithms, including AdaBoost, Decision Tree, Extra Tree, Random Forest, Naïve Bayes, K-Nearest Neighbour (KNN), and Neural Network, in predicting diverse levels of air quality. Each algorithm was assessed according to precision, recall, F1-score, and overall accuracy, with specific emphasis on difficult air quality classifications. The findings indicated that some models, including Decision Tree, Extra Tree, Random Forest, and Neural Network, attained excellent accuracy and F1-scores, whilst others, such as AdaBoost and Naïve Bayes, exhibited deficiencies in managing certain air quality categories. An ensemble model was created to address these constraints and improve overall forecast accuracy by integrating the capabilities of the most effective algorithms. The ensemble model exhibited outstanding performance, attaining flawless precision, recall, F1-scores, and accuracy across all air quality categories, signifying its potential as a highly dependable instrument for real-time air quality monitoring and prediction. This research finds that the ensemble model signifies a substantial improvement in air pollution forecasting. Therefore, providing an effective option for environmental monitoring systems. The research underscores the significance of amalgamating several machine learning algorithms to enhance model resilience and precision, offering critical insights for public health administration and policy formulation.