Browse Articles
Discover research articles across all indexed journals
Effect of using different fundus cameras and image resolutions on automatic measurements of retinal vascular parameters
Early prediction of wind turbine anomalies using 1D-CNN and temporal feature engineering on multi-source SCADA data
Abstract Early and accurate detection of anomalies in wind turbines is critical for ensuring system reliability, minimizing unplanned downtime, and reducing maintenance costs in large-scale renewable energy infrastructures. In this study, we propose a robust deep learning framework for wind turbine anomaly detection, leveraging a newly constructed dataset that integrates Supervisory Control and Data Acquisition (SCADA) data from three distinct wind farms. Extensive preprocessing and domain-specific temporal feature engineering were employed to capture complex patterns and enhance model generalizability across heterogeneous data sources. A comparative evaluation of several state-of-the-art deep learning models—including 1D Convolutional Neural Networks (1D-CNN), Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM), Bidirectional LSTM (Bi-LSTM), and Gated Recurrent Units (GRU); was conducted using standard classification metrics. Among these, the 1D-CNN consistently outperformed the recurrent models, achieving an accuracy and F1-score of 85%. This performance is attributed to the model’s capacity to effectively learn localized temporal dynamics in multivariate time series data. The findings demonstrate that a carefully designed 1D-CNN architecture, combined with strategic temporal feature engineering and multi-source data fusion, offers a scalable and accurate solution for early fault detection in wind turbine systems. This work lays the foundation for intelligent condition monitoring systems in the renewable energy sector. We then propose and evaluate a hybrid CNN-LSTM architecture augmented with an attention mechanism. This model leverages both CNN’s strength of extracting local features. And LSTM capacity to capture temporal dependencies, while the attention layer dynamically focuses on the most important segments of the sequence. Our findings show that the suggested hybrid model performs noticeably better than the independent base models, attaining higher generalization and accuracy. This work advances wind turbine fault detection through creating a diverse, multi-source wind farm dataset for superior generalizability; pioneering a reproducible benchmarking framework across deep learning models on heterogeneous data; and a hybrid CNN-LSTM with attention, surpassing baselines by 2% while enabling practical decision-making.
Perceived peer support and academic achievement among university students: the chain mediating roles of emotion regulation and behavioral engagement
Assessment of artificial intelligence-based control algorithms to be implemented in an affordable transradial myoelectric prosthesis
The most important features in generalized additive models might be groups of features
Feature reduction using swarm optimization and random forest classifiers for early diabetes risk prediction
The attractiveness of reclaimed and developed post-mining sites in Poland urban areas
Route evaluation strategy of the Beijing-Tianjin multi-airport system based on the two-dimensional evaluation framework
Pomiferin protects against sepsis-associated acute liver and kidney injury via inhibition of NF-κB activation, oxidative stress, and cytochrome-c
Dynamic channel allocation for secondary users in cognitive radio network
<tt>SeQuant</tt> framework for symbolic and numerical tensor algebra. I. Core capabilities
SeQuant is an open-source library for symbolic algebra of tensors over commutative (scalar) and non-commutative (operator) rings. The key innovation supporting most of its functionality is a graph-theoretic tensor network (TN) canonicalizer that can handle TNs with symmetries faster than their standard group-theoretic counterparts. The TN canonicalizer is used for the routine simplification of conventional tensor expressions, for optimizing the application of Wick’s theorem (used to canonicalize products of tensors over operator fields), and for the manipulation of the intermediate representation leading to the numerical evaluation. Notable features of SeQuant include support for noncovariant TNs (featuring hyperedges in their graphical representation and often arising from tensor decompositions) and for tensors with modes that depend parametrically on indices of other tensor modes (such dependencies between degrees of freedom are naturally viewed as nesting of tensors, or “tensors of tensors” arising in block-wise data compressions in data science and modern quantum simulation). SeQuant blurs the line between pure symbolic manipulation/code generation and numerical evaluation by including compiler-like components to optimize and directly interpret tensor expressions using external numerical tensor algebra frameworks. The SeQuant source code is available at https://github.com/ValeevGroup/SeQuant.
Development and validation of an AI use scale for sport and exercise science students
Abstract Artificial intelligence (AI) is rapidly transforming sport and exercise domains. Yet, sport-science curricula have lagged in integrating AI literacy, and little is known about students’ knowledge, ethical practices, and perceptions regarding AI. Existing measurement tools are often generic and ill-suited to sport education contexts. This study aimed to develop and validate a concise, domain-specific questionnaire to assess sport students’ AI use. A systematic instrument development process was used, including literature review, expert consultation, item generation, and psychometric validation. The resulting 14-item tool covers four domains: AI Awareness, Ethics & Disclosure, Trust & Verification, and Course & Institution Expectations. The instrument was administered to 864 undergraduate sport-science students in China and analysed using exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) on separate training and test sets. EFA supported a four-factor structure with high sampling adequacy (KMO = 0.95) and strong communalities. CFA confirmed good model fit (CFI = 1.00, TLI = 0.99, RMSEA = 0.09, SRMR = 0.05). Subscales demonstrated excellent internal consistency (Cronbach’s α = 0.90–0.94; McDonald’s ω = 0.90–0.94), convergent validity (AVE = 0.77–0.87), and discriminant validity (HTMT ratios < 0.85). This validated, context-specific instrument provides educators with a reliable tool to assess and enhance AI literacy in sport education. The findings support the integration of targeted AI training, ethical instruction, and institutional policies to prepare students for responsible, real-world AI engagement in sport and health domains.
Coarse-grained torsional potential of polypeptide backbone by fragment molecular orbital method: Dependence on coarse-graining scheme
Coarse graining of polymer chains involves a usually arbitrary assignment of atoms to extended sites. In this work, we use the scale-consistent theory of coarse graining, in which all-atom potential energy surfaces are mapped to the effective coarse-grained energy surfaces by partitioning the potential of mean force of a system into Kubo cluster-cumulant functions, to analyze the dependence of polypeptide-backbone coarse-grained torsional potential on the definition of coarse-grained sites. The polypeptide-backbone unit is modeled by a glycine residue and is divided into CH3–CONHCH2–CONHCH3 (CH3–BN1–BN2) or CH3CONH–CH2CONH–CH3 (BC1–BC2–CH3) sites, with methyl as a capping group. We use the fragment molecular orbital method to partition the all-atom potential energies into single- (site), two-body (site-pair), and three-body (site-triad) contributions to eliminate those involving the capping group. We demonstrate that the torsional potential corresponding to the BC1–BC2–CH3 partition differs remarkably from that of the CH3–BN1–BN2 partition and that obtained from the energy surface of whole terminally blocked glycine. This difference is caused by the electron-density leak outside the interacting BC1 and BC2 units, resulting from their borders running across the single N–C bonds. Consequently, a preferable choice of the boundary between connected sites is to set it across a bond between atoms with a similar electronegativity.
Elucidation of vaginal microbiota of women associated with bacterial vaginosis from Northern region of India
New range-separated screened and full-range hybrid functionals
We develop and assess new full-range and range-separated (RS) hybrid density functionals combining G96 or WC exchange with Perdew–Burke–Ernzerhof (PBE) and Lee–Yang–Parr correlations. The B1-type full hybrids, generalized through a Puiseux-series expansion of the adiabatic connection, were benchmarked against the Ghosh–Oshi–Salahub-0 database spanning main-group and transition-metal thermochemistry. Among full hybrids, only PBE0 improves upon its parent generalized gradient approximation, while others overcorrect. RS hybrids achieve balanced and transferable accuracy; GSG2 (MAE ≈ 5.97 kcal mol−1) slightly outperforms GS2 and HSE06. These results underscore that range separation is essential for broad applicability, establishing the Ghosh–Salahub–Gill and Ghosh–Salahub families as versatile screened hybrid functionals with promising utility in both molecular and condensed-phase systems.
Toward accelerating fluvial morphodynamic simulations through a speed accuracy trade-off assessment
Mechanical and electrical anharmonicity in the infrared spectra of graphene fragments: From planar polycyclic aromatic hydrocarbons to buckybowls
Polycyclic aromatic hydrocarbons (PAHs) and their curved buckybowl derivatives are key motifs in combustion, atmospheric, and astrochemical environments, as well as fundamental building blocks of graphene-based materials. Here, we show that the Pisa composite schemes provide an accurate and affordable framework for predicting equilibrium structures, rotational constants, and vibrational spectra across representative planar (indene and azulene) and curved (corannulene and sumanene) PAHs. Accurate equilibrium geometries are combined with a fully anharmonic treatment based on generalized second-order vibrational perturbation theory, including both mechanical and electrical contributions to infrared intensities. This enables reliable prediction not only of fundamental frequencies but also of overtone and combination-band intensities, which are strictly absent in the harmonic approximation. Indene and azulene validate the structural and spectroscopic accuracy of the protocol for planar π-systems, whereas corannulene and sumanene quantify the effects of bowl curvature on mode mixing, symmetry breaking, and activation of otherwise forbidden transitions. Across all systems, the proposed computational approach yields fundamental frequencies with typical deviations of 5–10 cm−1, reproduces curvature-dependent shifts in C–H stretching and ring-deformation modes, and captures characteristic splitting patterns and intensity redistribution induced by both mechanical and electrical anharmonicity.
Single-nucleus ATAC-seq analysis resolves chromatin and transcriptional features of fibrolamellar carcinoma
Ab initio mixed polarizabilities responsible for the Raman-forbidden <i>ν</i> 2 transition in carbon dioxide molecule
The vibrational parity of a centrosymmetric molecule can be broken in the Raman spectra when the electric-dipole (E1) virtual transitions are accompanied by the electric-quadrupole (E2) or magnetic-dipole (M1) ones. To assess the intensity of the not yet detected ν2 CO2 Raman band, accurate ab initio values of the mixed E1–E2 (Â) and E1–M1 (Ĝ) polarizabilities of a bent CO2 molecule are presently obtained using the DALTON program suite. The previous SCF results of Amos, Buckingham, and Williams [Mol. Phys. 39, 819 (1980)] for the Â-tensor components are substantially refined by employing the extended basis sets and accounting for the electron correlation (up to CCSD). By using the SOPPA approach, the derivatives of Ĝ are obtained for the first time. The results are processed to derive the rank-r irreducible spherical tensors (IST), Xκ(r), which vary in space as the Wigner functions Dκm(r) whose subscript κ determines the dependence on the bent-molecule plane orientation relative to the symmetry axis. As a result, four ISTs (A1 (1), A1 (2), A1 (3), and A3 (3)) fully determine the ν2 E1-E2 intensity whereas tensors G1 (1) and G1 (2) do the same for the magnetic channel. Using the thus derived properties, the integrated ν2 band intensities are calculated for the first time and, like our recent assessments for the ν3 band [Kouzov et al., Opt. Spectrosc. 133, 581 (2025)], show the leading role of the E1–M1 terms. The results could be helpful for detection of this novel parity-changing Raman process.
A computational rule-based model of MAPK/ERK system regulation
Abstract The MAPK/ERK pathway coordinates multiple cellular functions, including proliferation, apoptosis, and motility, yet it is frequently modeled as a one-purpose cascade. Pathway complexity stems from the existence of isoforms that are regulated differently and have specific interacting partners. Here, we overcome combinatorial complexity by constructing a rule-based model of the MAPK/ERK pathway that accounts for differential regulation of MEK and RAF isoforms and RAF interactions with 14-3-3 proteins. The model addresses signaling based on the enzymatic cascade as well as regulatory protein-protein interactions. We propose that at low concentrations of growth factors, RAS is activated only in a portion of the membrane. This allows the model to reconcile the observed graded and switch-like responses observed at the upper (RAS and RAF) and the lowest (ERK) tiers of the pathway, respectively. We demonstrate that functional differences between BRAF and CRAF, or ARAF, can arise from their distinct interactions with 14-3-3. The 14-3-3 dimers inhibit all RAF monomers in the closed form and preferentially stabilize BRAF-CRAF or BRAF-ARAF dimers and may not stabilize RAF dimers without BRAF. The constructed model is used to explore qualitative differences in MAPK/ERK pathway signaling across different cell lines. The exploration starts from the nominal model parameters chosen to reflect a generic cell.