Search Results (202)

Background and Objectives: Pain has been proposed as a pathway linking chronic disease to healthcare spending, but existing estimates measure exposure, pain and cost in the same period and combine coefficients on incompatible scales. We asked how much of the disease–expenditure association runs through pain when exposure, pain and cost are separated in time. Materials and Methods: In the Korea Health Panel Survey (2019–2023; 54,845 adult person-years), chronic disease at year t, EQ-5D pain/discomfort at t + 1 and out-of-pocket payments during t + 2 were linked, conditioning on pain and payments at t. Direct and indirect effects were estimated by parametric g-computation (randomised-interventional analogues, two-part outcome model, exposure-specific adjustment sets, survey and censoring weights, false discovery rate control). Results: Among 25,692 triplets, pain/discomfort predicted the amount spent among healthcare users (cost ratio, CR 1.099, 95% confidence interval, CI 1.023–1.180) but showed no detectable association with whether care was used (odds ratio, OR 1.136, 0.918–1.406); the contemporaneous cost ratio was substantially larger (1.296), inflated by simultaneity. Three musculoskeletal conditions prospectively predicted pain (false discovery rate q < 0.001). Six conditions had indirect-effect intervals excluding zero, but none survived false discovery rate correction (minimum q = 0.19); where estimable, the proportion mediated was small (2.8% to 4.8%). Conclusions: The disease → pain and pain → payment associations are each established, but the disease-specific mediated amount is not confirmed after multiplicity correction. Contemporaneous designs overstate the indirect association; claims that pain causes a specified share of chronic disease expenditure are not supported. Full article

Objectives: We evaluated the performance of mammography-based radiomics in prediction of clinical nodal status, clinical tumor stage, clinical disease stage, status of the PIK3CA mutation and concordance with the radiologist-assessed BI-RADS category. Mammography is often the first imaging modality performed in the screening or diagnostic workup of breast cancer. Materials and Methods: In this single-center retrospective study, we included 102 histopathologically confirmed cases of breast cancer from 100 patients. The tumor region and whole-breast parenchyma were contoured on 368 craniocaudal and mediolateral-oblique mammograms. Cases were temporally split into training and test sets by histopathological diagnosis date (training: 70%, held-out test: 30%). Six linear models were tuned by 5-fold x 3-repeat stratified cross-validation. The winning model per endpoint was applied once to the test set. We report AUCs with 95% confidence intervals, permutation p-values and Benjamini–Hochberg false discovery rate (BH-FDR) correction across five endpoints. Results: Three of the studied endpoints reached significance after BH-FDR correction: clinical nodal status (cN0 vs. cN-positive; AUC: 0.726 (95% CI: 0.501–0.886)), clinical tumor stage (cT1-2 vs. cT3-4; AUC: 0.792 (95% CI: 0.575–0.929)) and overall disease stage (I–II vs. III–IV; AUC: 0.778 (95% CI: 0.572–0.917)). Mammography-based radiomics failed to predict the presence of PIK3CA mutation and concordance with radiologist-assessed BI-RADS category. Conclusions: Mammography-based radiomics has shown a hypothesis-generating discriminative value for differentiation between cN-negative and cN-positive disease, early from advanced clinical tumor stage, and early from advanced overall disease stage in breast cancer. Mammography-based radiomics did not predict PIK3CA status and did not discriminate between radiologist-assessed BI-RADS 4 and 5 groups. Full article

Background and Objectives: Psychiatric symptoms and multidimensional impulsivity may be associated with the clinical and functional heterogeneity of substance use disorders (SUDs), but evidence from Saudi Arabian treatment populations remains limited. This study examined concurrent associations with a study-specific aggregate of Addiction Severity Index (ASI-5) interviewer severity ratings. Materials and Methods: This cross-sectional study included 204 adults receiving inpatient detoxification or residential rehabilitation services in Najran, Saudi Arabia. The seven ASI interviewer ratings were summed as an aggregate score. Psychiatric symptoms were assessed using the Brief Psychiatric Rating Scale (BPRS), and impulsiveness using the Barratt Impulsiveness Scale-11 (BIS-11). Analyses included age- and sex-adjusted analyses of covariance, false-discovery-rate correction, bootstrap Spearman correlations, an a priori multivariable model with HC3 robust confidence intervals, and sensitivity analyses addressing content overlap, multicollinearity, influential observations, functional form, and the bounded outcome distribution. Results: The mean age was 34.4 ± 7.3 years; 78.9% were male. The aggregate score was 26.33 ± 4.15. It correlated strongly with BPRS total (ρ = 0.874, 95% CI 0.827–0.908) and moderately with BIS-11 total (ρ = 0.590, 95% CI 0.494–0.669). In the primary model (adjusted R² = 0.864), age, BPRS total, motor impulsiveness, and non-planning impulsiveness were conditional correlates of greater aggregate burden; male sex was inversely associated, and polysubstance use was not. BPRS remained associated after exclusion of the ASI psychiatric domain. Separate impulsivity-domain and BIS-11 total models, 5000-resample bootstrap analysis, median regression, and fractional-logit analysis retained the principal psychiatric and impulsivity directions. The polysubstance coefficient varied across sensitivity models. Conclusions: Psychiatric symptom burden and selected impulsivity dimensions were concurrently associated with higher study-specific aggregate ASI interviewer severity scores. Integrated psychiatric and impulsivity assessment may help characterize patients with greater concurrent multidomain clinical burden; the aggregate score should not be interpreted as a validated global severity or prediction measure. Full article

Hydrogen (H₂) is expected to play a critical role in modern industry, particularly in ammonia synthesis, petroleum refining, and low-carbon transportation. The safe storage of H₂ remains a major challenge due to its low volumetric density under ambient conditions. Metal–Organic Frameworks (MOFs), highly porous crystalline materials, have emerged as promising H₂ storage candidates owing to their high surface areas and tuneable pore structures. Molecular simulations such as grand canonical Monte Carlo or density functional theory are costly and limited in exploring large material spaces, motivating efficient predictive tools to accelerate discovery. Here, Machine Learning (ML) techniques are compared to an explainable artificial intelligence (XAI) approach using symbolic regression (SR), trained on 10,123 experimentally measured H₂ adsorption datapoints from real-world MOFs. The best performing model achieved a goodness of fit of 0.9986 with lower computational demand, but reduced interpretability, addressed using XAI analysis and clustering. SR achieves a lower goodness of fit of 0.914 but produces a physically meaningful equation highlighting structural features driving high gravimetric efficiencies. These results demonstrate strong ML capability for predicting how MOF properties and environmental conditions affect H₂ uptake. This offers engineers and researchers a practical means of screening potential MOFs for H₂ storage applications, with the XAI analyses providing additional confidence in the predictions. They allow researchers to understand the physical reasoning behind each output, assess the reliability of individual predictions, and make fully informed decisions, enabling predictive models to be acted upon with confidence in real-world contexts. Full article

Driven by advances in renewable energy technologies, research on perovskite optoelectronics has advanced rapidly across material exploration, device engineering, and intelligent integrated systems. Conventional trial-and-error experiments face inherent constraints in precisely regulating perovskite chemical compositions and microstructures, as well as in mitigating degradation in perovskite solar cells (PSCs). Artificial intelligence (AI) and the Internet of Things (IoT) have emerged as powerful tools for material discovery, synthetic condition design, and the prediction of perovskite fundamental properties and device outputs. This review systematically summarizes recent advances in machine learning (ML) implementations for PSC research, covering molecular-scale material screening, synthetic parameter optimization, performance forecasting, device architecture design, and system performance evaluation. We further elaborate on key obstacles hindering ML-assisted perovskite development, including insufficient operational stability, barriers to large-scale fabrication, and limited computational efficiency. Last, we outline promising research avenues and highlight the transformative capacity of ML to advance high-performance, manufacturable perovskite optoelectronic devices. Full article

Artificial intelligence (AI) agents and large language model (LLM) agents are beginning to move materials discovery beyond isolated prediction tasks and toward tool-grounded workflows that can retrieve prior knowledge, configure simulations, launch calculations, inspect outputs, and decide what to do next. However, adjacent reviews on materials informatics, self-driving laboratories, natural-language processing in materials science, and autonomous chemistry have not isolated simulation-driven materials workflows as a distinct evidence base. This review addresses that gap through PRISMA-guided searches in Scopus (8 May 2026) and Web of Science (15 June 2026) for English-language journal articles published between 2022 and 2026. The combined search returned 232 records; 27 full texts were assessed and 26 studies were included in the final qualitative synthesis after one full-text exclusion. No eligible study was published in 2022 or 2023, indicating that the field emerged only in 2024 and expanded rapidly in 2025–2026. Catalysis and adsorption tasks (n = 6) and alloy design or evaluation (n = 5) dominated the corpus, while specialized multi-agent architectures were the most common pattern (n = 13). Across the included studies, agentic reasoning was most often coupled to workflow orchestration or integration tools, molecular-dynamics or atomistic simulation environments, and materials-data or machine learning screening pipelines; public repositories or archival artifacts were reported in 18 of 26 studies, experimental validation in six, and robotic closed-loop execution in only one study. The strongest evidence came from workflows that grounded language model decisions in simulators, structured databases, or experimentally verifiable outputs rather than in free-form text alone. This review therefore establishes AI agent workflow orchestration as a distinct analytical category within materials discovery and identifies the reporting, validation, and reproducibility conditions required for these systems to function as credible scientific infrastructure rather than as conversational demonstrations. Full article

Material screening, including oxide catalyst discovery, remains constrained by costly synthesis-and-test cycles, making sample-efficient experiment selection a central challenge. Active learning can accelerate this process, but standard acquisition rules mainly optimize predictive utility and offer limited support for clarifying the causal structure behind material-screening targets. CIAL is a causal-structure-aware acquisition framework. It combines an intervention-inspired graph-structure term with expected improvement, using an adaptive schedule that begins with structure clarification and shifts toward prediction-oriented exploitation. On synthetic benchmarks with known ground-truth causal graphs, CIAL reduces final-iteration structural Hamming distance by 76.2% relative to the standard EI baseline; significance is assessed using two-sided Welch’s t-tests on final-iteration metrics across five random seeds, with $p=0.0036$ for SHD and $p<0.001$ for F1. For real-data screening, CIAL is evaluated on a Materials Project transition metal oxide benchmark built from filtered oxide entries with formation-energy labels and generic bulk/compositional descriptors. Formation energy is an upstream stability-relevant property, not a direct catalytic activity label. On the main MP benchmark, CIAL is competitive with EI; its clearest advantage over EI appears under NOTEARS-based cross-split validation, but random sampling achieves the highest inferred-reference graph agreement in several real-data settings. An exploratory OC20 catalyst dataset check follows the same ordering but remains statistically inconclusive. Ablation results show that neither the causal nor the predictive component alone recovers the full benefit of the combined policy. CIAL improves graph recovery on controlled synthetic benchmarks; the real-data evidence is mixed and protocol-dependent, establishing CIAL as a proof-of-concept structure-aware acquisition method rather than a validated catalyst-discovery framework. Full article

Lead-free double-absorber perovskite solar cells offer broad-spectrum absorption and environmental benefits, but their multilayer heterostructure creates computational challenges for conventional design optimization. This study introduces an automated framework integrating SCAPS-1D simulation, multilayer perceptron (MLP) surrogate modeling, metaheuristic algorithms, and reinforcement learning (RL). Using FTO/ZnO/Cs₂TiBr₆/RbGeI₃/CuI/Au cells, the MLP model trained on Latin hypercube sampling data achieved high accuracy (R² > 0.95). The proximal policy optimization (PPO) RL agent converged to 27.41% power conversion efficiency (PCE) in approximately 20 steps. For direct 15-dimensional optimization, simulated annealing and particle swarm optimization reached 98% target PCE with 138 and 111 function evaluations, respectively, while Grey Wolf Optimizer (GWO) yielded the highest average PCE. Transfer learning successfully adapted the pretrained model to a novel FASnI₃/Sb₂S₃ structure, improving the prediction accuracy of PCE, J_SC, and FF. This work systematically optimizes Cs₂TiBr₆/RbGeI₃ solar cells while establishing an efficient, generalizable paradigm for intelligent photovoltaic device design, validation, and material discovery. Full article

Background: Metallic orthopedic implants face persistent clinical challenges that have proved resistant to incremental conventional development. Machine learning and artificial intelligence offer a complementary paradigm for navigating the high-dimensional design spaces governing implant performance, yet the literature remains fragmented across disciplinary silos with no comprehensive synthesis spanning the full development pipeline. Methods: A structured database search of PubMed/MEDLINE, Embase, and Cochrane (executed May 2026), supplemented by hand-searching of reference lists, identified 33 primary studies organized across five sequential domains: alloy composition discovery, additive manufacturing process–property optimization, lattice and porous structure design, surface engineering and coatings, and corrosion and wear prediction. Results: Across all five domains, machine learning approaches, including random forests, convolutional neural networks, Bayesian optimization, generative adversarial networks, physics-informed neural networks, and autonomous multi-agent platforms, have accelerated property prediction and design space exploration beyond experimental or simulation-based methods. Shared barriers to translation include small, heterogeneous datasets, reliance on internal rather than external validation, limited interpretability, and the absence of regulatory frameworks for AI-assisted device design. Representative performance included modulus predictions within ~4 GPa of first-principles values, ML-designed alloys reaching ~42.7 GPa (versus 103–120 GPa for Ti-6Al-4V), property prediction R² often above 0.90 (up to 0.96–0.9991), 98.3% corrosion severity classification accuracy, and acceleration from a roughly fivefold reduction in finite element simulations to surrogates compressing days into minutes. Conclusions: Addressing these limitations will require open standardized databases linking materials parameters to registry-level clinical outcomes, prospective clinical validation studies, and coordinated engagement between researchers, industry, and regulatory agencies. Full article

Chemical artificial intelligence increasingly relies on molecular property prediction models trained from proprietary compound libraries, bioassay records, and reaction-screening data. However, these data often contain commercially sensitive structures, confidential activity labels, and privacy-relevant experimental metadata, making direct centralization impractical. This paper proposes PrivMol, a privacy-preserving computational chemistry framework for federated molecular representation learning. PrivMol introduces two novel algorithms: Secure Substructure-Aware Federated Optimization and Differentially Private Molecular Gradient Calibration. The first algorithm decomposes molecular graphs into privacy-sensitive and task-relevant substructure regions, enabling local clients to train graph neural networks while transmitting only securely aggregated model updates. The second algorithm adaptively calibrates clipping and perturbation according to atom- and substructure-level contribution scores, reducing unnecessary utility loss on chemically informative fragments while retaining formal differential privacy guarantees. To improve robustness under heterogeneous chemical spaces, PrivMol incorporates local contrastive molecular alignment without exposing raw molecules, labels, scaffolds, substructure masks, or embeddings. Experimental evaluation on widely used public molecular benchmarks, including ESOL, FreeSolv, Lipophilicity, BBBP, BACE, HIV, and Tox21, demonstrates that PrivMol provides a favorable trade-off among prediction accuracy, communication efficiency, empirical leakage resistance, and privacy protection. The study offers a practical route toward secure collaborative chemical intelligence for computer-aided drug discovery, toxicology prediction, and materials informatics. Full article

Background and Objectives: The clinical differentiation of peripheral involvement within the spondyloarthritis (SpA) spectrum remains a significant challenge. Identifying patients at the stage of psoriasis without clinical arthritis (PsO), before the onset of overt arthritis, is crucial for early disease management. MicroRNAs (miRNAs) have emerged as potential epigenetic regulators in inflammatory rheumatic diseases. This study aimed to identify circulating miRNA profiles that serve as discriminative biomarkers between PsO and peripheral SpA (p-SpA). Materials and Methods: This exploratory study was conducted in two phases. In the preliminary discovery phase, plasma miRNA expression was analyzed using high-throughput microarrays in patients with p-SpA (modeled by peripheral psoriatic arthritis, n = 6), PsO (psoriasis without clinical arthritis, n = 3), and osteoarthritis (n = 3). In the validation phase, candidate miRNAs were verified using TaqMan-based qPCR in an independent cohort (n = 30) of various SpA phenotypes, categorized into those with peripheral arthritis (SpA-A) and those without (SpA-N). Results: The preliminary discovery phase identified altered levels of hsa-miR-140-5p, hsa-miR-192-5p, and hsa-miR-146a-5p among the groups; however, due to the small sample size, these initial findings required strict downstream verification. Functional enrichment analysis revealed that these candidate miRNAs primarily targeted the NF-κB signaling pathway (hsa04064) and Toll-like receptor (TLR) signaling pathway (hsa04620). The validation cohort confirmed that these three miRNAs could reliably differentiate SpA-A from SpA-N patients. Furthermore, bioinformatic mapping predicted that downstream targets, including TRAF6, IRAK1, and CXCL2, may be associated with these clinical phenotypes, serving as hypothesis-generating observations for future studies. Conclusions: Our findings suggest that specific plasma miRNA profiles are associated with the inflammatory pathways driving peripheral involvement in the SpA spectrum. These miRNAs represent potential biomarkers associated with peripheral arthritis in the SpA spectrum. While they offer new molecular insights into disease pathogenesis, their predictive value for identifying PsO patients at risk of developing overt arthritis requires confirmation in future longitudinal studies. Full article

Cu-Al-Ni shape memory alloys (SMAs) are promising solid–solid phase-change materials (PCMs) for transient thermal management. Data-driven screening for high-latent-heat (ΔH) Cu-Al-Ni PCMs across the vast compositional space is efficient, but predictive accuracy and screening reliability degrade when noisy experimental data are used. A noise-aware machine learning strategy was applied to accelerate the discovery of high-ΔH Cu-Al-Ni alloys with martensite start temperature (M_s) within the 100–200 °C range from noisy experimental datasets. The optimal noise level was estimated by minimizing the prediction error of the noise-aware Kriging model. The application of this strategy led to the discovery of four Cu-Al-Ni alloys with M_s ranging from 125 to 163 °C and ΔH ranging from 9.27 to 9.86 J/g. The best-performing Cu₈₄Al₁₃Ni₃ (wt.%) alloy achieved M_s = 163 °C, ΔH = 9.86 J/g, thermal conductivity of 102 W·m⁻¹·K⁻¹ and figure of merit of 7272 × 10⁶ J² K⁻¹ s⁻¹ m⁻⁴. Its ΔH exceeds the previous highest Cu-Al-Ni ΔH in the 100–200 °C window by 11.8%, while its FOM exceeds the previous highest Cu-Al-Ni FOM by 33.75% and represents the highest value among the surveyed PCMs within the 100–200 °C range. After 100 thermal cycles, ΔH decreased by 0.158 J/g and M_s shifted by 0.9 °C, demonstrating good thermal cycling stability. Full article

The ceramic and refractory industries are undergoing a progressive transition toward more sustainable and resource-efficient manufacturing systems driven by increasing environmental regulations, rising energy demands, and the need to reduce dependence on virgin raw materials. In this context, artificial intelligence (AI) has emerged as a promising tool for improving material design, process optimization, predictive maintenance, and circular manufacturing strategies. This review provides a comprehensive analysis of recent advances in AI applications within ceramic and refractory systems, with particular emphasis on their role in enabling circular economy approaches and intelligent manufacturing environments. The study examines the integration of machine learning, deep learning, computer vision, digital twins, and Industry 4.0 technologies across multiple domains, including materials discovery, defect detection, waste classification, process control, and sustainability assessment. In addition, the review discusses the incorporation of secondary raw materials such as fly ash, slag, waste glass, ceramic sludge, and spent refractories into circular ceramic production systems. The analysis highlights the potential of AI-driven methodologies to improve resource efficiency, reduce environmental impact, and enhance process adaptability under complex industrial conditions. Furthermore, current limitations associated with data availability, model interpretability, industrial scalability, and integration with life cycle assessment frameworks are critically discussed. Finally, future research directions are identified, emphasizing the development of standardized datasets, hybrid experimental–AI methodologies, digital manufacturing ecosystems, and intelligent decision-making systems for next-generation sustainable ceramic and refractory technologies. Full article

Phenomenological crystal plasticity (CP) models are widely used in Integrated Computational Materials Engineering (ICME) to bridge microstructural features with engineering-scale mechanical behaviour. However, their practical application is hindered by two major challenges: high computational costs of physics-based simulations, and the labour-intensive, trial-and-error nature of parameter calibration. These challenges are amplified in additively manufactured (AM) materials, where location-dependent properties require calibration to be repeated at multiple points to produce a detailed property map. Additionally, a limited understanding of how individual parameters of the CP models influence stress–strain predictions across the strain spectrum compounds these issues, making it challenging to utilise CP models for efficient materials design. To address these limitations, we developed an integrated framework that combines deep neural network (DNN) surrogates, optimisation algorithms (OAs), and explainable AI (XAI) techniques. We also utilised experimental tensile data from AM AlSi10Mg alloy as ground truth since AM materials are expected to benefit the most from our investigation. We demonstrate that, by using OAs such as a Natural Evolutionary Strategy or a Genetic Algorithm, the calibration process can be made more accurate and significantly accelerated. We also investigated the utility of employing deep neural network (DNN) surrogates of CP simulations in the calibration process. The fast-solving DNN surrogates achieved substantial time savings in the absence of OAs, i.e., during exhaustive parameter searches mandated by trial-and-error strategies. However, their effectiveness in parameter discovery was context-dependent when used in conjunction with OAs, since OAs can sometimes converge with fewer simulations than required for DNN training. Furthermore, we applied Shapley Additive exPlanations (SHAP), an XAI method, which revealed intricate interactions among some CP parameters, offering insight into why conventional trial-and-error calibration approaches often prove challenging. Our study contributes to strengthening the practical relevance of CP models for modelling-informed materials engineering and optimisation applications. Finally, our integrated framework offers broad applicability beyond materials modelling, enabling accelerated discovery of tuneable parameters in phenomenological models and providing deeper insight into their contributions to predictions. Full article

The integration of artificial intelligence (AI), machine learning (ML), and computational modeling with experimental catalysis is reshaping materials design and chemical process development. Tailored heterogeneous catalysts including supported metals, zeolites, defect-engineered materials, and multi-element systems exhibit enhanced activity, selectivity, and stability through engineered active sites and porosity. AI and ML approaches enable predictive modeling, high-throughput screening, mechanistic insight, and rational catalyst design by linking synthesis conditions, structural features, and performance metrics across scales. Applications span CO₂ conversion, methane reforming, hydrogen production, polymer recycling, and photocatalysis, with platforms such as PHOTOREAC, QMOF, and PhotoCatDB facilitating the translation from laboratory experiments to reactor-scale processes. Hybrid strategies that combine mechanistic understanding with data-driven models improve interpretability, predictive accuracy, and process optimization. These advances underscore a paradigm shift toward data-driven catalysis, accelerating discovery, supporting sustainable chemical technologies, and emphasizing the role of human expertise in guiding responsible AI deployment. Full article