Present and future mobile networks combine wireless radio access technologies from multiple cellular network generations, all of which coexist. Seamless Vertical Handover (VH) decision-making is still a challenging issue in heterogeneous cellular networks due to the dynamic conditions of networks, different demands on QoS, and the latency of the handover process. Maintaining a very high-accuracy VH decision requires considering several network parameters. There is a trade-off between the gain of the VH accuracy and the corresponding latency in the computational complexity of the decision-making methods. This paper proposes a lightweight VH prediction DL strategy for 3G, 4G, and 5G networks based on the Light-Gradient Boosting Machine (LGBM) feature selection and Peephole Long Short-Term Memory (PLSTM) prediction model. For dense networks with large datasets and high-dimensional data, the combination of PLSTM and the fast feature selection LGBM, can reduce the computing complexity while preserving prediction accuracy and excellent performance levels. The proposed methods are evaluated using three case study scenarios using different feature selection thresholds. The performance evaluation is achieved by training and testing the proposed model, which shows an improvement using the proposed LGBM and PLSTM in terms of reducing the number of features by 64.28% and enhancing the VH accuracy prediction by 43.81% in Root Mean Squared Error (RMSE), and reducing the VH decision time of up to 51%. Furthermore, a network simulation using the proposed VH prediction algorithm shows an enhancement in overall network performance, with the number of successful VHs being 87%. Consequently, the data throughput is significantly enhanced.

Transaction fees play a crucial role in determining the efficiency and scalability of blockchain networks, particularly in Ethereum, where gas fees fluctuate significantly due to network congestion and competitive bidding. This study analyzes transaction fee patterns in the Ethereum blockchain and their impact on network efficiency by examining key blockchain metrics such as block density, transaction size, and transaction fee variability. The findings indicate that the mean transaction fee is 0.0342 ETH, with a median of 0.0008 ETH, demonstrating significant fee variability. The study also finds a strong positive correlation (r ≈ 0.75, p < 0.01) between transaction fees and block density, as well as a moderate correlation with transaction size (r ≈ 0.58, p < 0.01), highlighting the direct impact of network congestion on fee structures. Time series forecasting with Autoregressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM) models reveals cyclical trends in transaction fees, often influenced by major network activities such as NFT releases, DeFi protocol surges, and high-frequency trading. The LSTM model achieves a lower RMSE (0.09) compared to ARIMA (0.15), demonstrating its superior predictive capability for fee trends. Additionally, anomaly detection techniques identify outlier transactions with fees exceeding 2.5 ETH, often associated with front-running strategies, priority gas auctions (PGA), and inefficient smart contract executions. Despite improvements introduced by EIP-1559, the findings indicate that Ethereum’s transaction fee market remains highly volatile, with block density fluctuating between 512.0% and 3896.0%, causing extreme fee spikes during congestion periods. The presence of large transactions (maximum size: 250 bytes) further amplifies fee inefficiencies, reinforcing the need for improved scalability solutions. This study underscores the necessity of Layer-2 rollups, dynamic block size adjustments, and more adaptive fee mechanisms to enhance blockchain efficiency. Future research should explore comparative studies across blockchain networks, advanced predictive modeling techniques, and the role of miner extractable value (MEV) in transaction ordering fairness. The study’s insights provide valuable guidance for developers, users, and policymakers aiming to optimize Ethereum’s transaction fee structure and enhance overall blockchain performance.

Advanced techniques for electricity consumption prediction in buildings using comparative correlation analysis, data normalization, and Long Short-Term Memory (LSTM) networks: A case study of a U.S. commercial building

Nondestructive detection of biogenic amines in muscle of Chinese mitten crab (Eriocheir sinensis): A basis for quality assessment using infrared spectroscopy and deep learning.

A robust hybrid machine learning framework for short-term load forecasting: integrating multi-linear regression, long short-term memory, and feed-forward neural networks for enhanced accuracy and efficiency

Microgel double-crosslinked hydrogel with excellent mechanical properties for flexible electronics.

This systematic review evaluates artificial intelligence (AI)-based predictive models developed for early sepsis detection in adult hospitalized patients. It explores model types, input features, validation strategies, performance metrics, clinical integration, and implementation challenges. A systematic search was conducted across PubMed, Scopus, Web of Science, Google Scholar, and CENTRAL for studies published between January 2015 and March 2025. Eligible studies included those developing or validating AI models for adult inpatient sepsis prediction using electronic health record data and reporting at least one performance metric (area under the curve [AUC], sensitivity, specificity, or F1 score). Studies focusing on pediatric populations, lacking quantitative evaluation, or unpublished in peer-reviewed journals were excluded. Data extraction followed preferred reporting items for systematic reviews and meta-analyses guidelines. Extracted variables included study design, patient population, model type, input features, validation approach, and performance outcomes. A total of 52 studies met the inclusion criteria. Most used retrospective designs, with limited prospective or real-time clinical validation. Commonly used algorithms included random forests, neural networks, support vector machines, and deep learning architectures (long short-term memory, convolutional neural network). Input data varied from structured sources (vital signs, laboratory values, demographics) to unstructured clinical notes processed via natural language processing. Reported AUC values ranged from 0.79 to 0.96, indicating strong predictive performance across models. AI models demonstrate significant promise for early sepsis detection, outperforming conventional scoring systems in many cases. However, generalizability, interpretability, and clinical implementation remain major challenges. Future research should emphasize externally validated, explainable, and scalable AI solutions integrated into real-time clinical workflows.

Pre-Implantation factor (PIF) restores working memory and promotes microglial ramification in the adult Dp(16)1Yey mouse model of Down syndrome.

Fibromyalgia is characterized by widespread musculoskeletal pain, fatigue, disturbances in cognitive and emotional functioning. Cognitive impairment (so-called "fibrofog") is increasingly recognized as a central feature of fibromyalgia. However, its relationship with psychological traits and personality dimensions remains insufficiently explored. This study aimed to examine the interaction between cognitive performance, emotional states, and personality traits in individuals diagnosed with fibromyalgia. This descriptive pilot study involved 10 female outpatients diagnosed with fibromyalgia according to the criteria of the American College of Rheumatology. Participants underwent a comprehensive assessment including neuropsychological, psychological, personality, and motor evaluations. Instruments used included the Young Schema Questionnaire, the State-Trait Anxiety Inventory, the Beck Depression Inventory, the Brief COPE questionnaire, and a battery of cognitive and motor tests. Statistical analyses were conducted using Spearman's rank correlation coefficient. The most frequently observed maladaptive personality schemas were self-sacrifice (80%) and unrelenting standards (70%). Trait anxiety was present in 80% of participants, and depressive symptoms were reported by 90%. While short-term and long-term memory were generally preserved, 80% of participants showed deficits in divided attention, and 40% demonstrated impaired mobility under dual-task conditions. Significant correlations were found between anxiety and cognitive flexibility, as well as between coping strategies and working memory performance. Patients with fibromyalgia exhibit a complex cognitive-emotional profile characterized by attentional deficits, maladaptive personality traits, and elevated psychological distress. These findings emphasize the importance of multidimensional assessment and suggest that interventions targeting personality schemas and coping mechanisms may improve cognitive and functional outcomes in this population.

Mild hyperhomocysteinemia alters spatial and recognition memories in male, but not female rats. Are inflammation, blood-brain barrier damage and Tau expression sex-specific predictors?

PreRBP: Interpretable deep learning for RNA-protein binding site prediction with attention mechanism.

Modeling transient temperature in phase change materials using a hybrid convolutional neural network and long short-term memory approach for melting process analysis

A data-driven architecture fusing Kolmogorov-Arnold feature extraction and contextual-attention long short-term memory network for accurate state-of-charge estimation in lithium-ion batteries under dynamic operating conditions

A neuromorphic photodetector with ferroelectric-controlled static, event, and short-term memory modes for on-chip real-time spatiotemporal classification and motion prediction

Construction of a machine learning-based prediction model for rice varieties-cooking-mastication.

An important component for the dependable and safe utilization of lithium-ion batteries is the ability to accurately and efficiently predict their remaining useful life (RUL). In this research, a long short-term memory recurrent neural network (LSTM RNN) model is trained to learn from sequential data on discharge capacities across different cycles and voltages. The model is also designed to function as a cycle life predictor for battery cells that have been cycled under varying conditions. By leveraging experimental data from the NASA battery dataset, the model achieves a promising level of prediction accuracy on test sets consisting of approximately 200 samples.

Wheat yield forecasts with seasonal climate models and long short-term memory networks

Self-attention based Difference Long Short-Term Memory Network for Industrial Data-driven Modeling

CCLR-DL: A novel statistics and deep learning hybrid method for feature selection and forecasting healthcare demand.

Anomaly detection in offshore open radio access network using long short-term memory models on a novel artificial intelligence-driven cloud-native data platform