Search Results (889)

Systemic sclerosis (SSc) is a complex autoimmune connective tissue disease characterized by microvascular dysfunction, immune activation, and progressive fibrosis affecting multiple organs, including the heart. Myocardial involvement represents an important but frequently underrecognized manifestation of SSc and may develop even in the absence of overt clinical symptoms. Cardiac manifestations include ventricular dysfunction, arrhythmias, conduction abnormalities, and heart failure, contributing substantially to morbidity and mortality. The underlying pathophysiology involves coronary microvascular dysfunction, immune-mediated myocardial inflammation, and progressive myocardial fibrosis, which often precede clinically apparent cardiac disease. This review aims to summarize the current understanding of myocardial involvement in SSc and to provide a comprehensive overview of contemporary multimodality cardiovascular imaging techniques for its detection, characterization, and risk stratification. A comprehensive overview of the current literature was conducted focusing on established and emerging cardiovascular imaging modalities for the evaluation of myocardial involvement in SSc. Particular attention was given to echocardiography, cardiac magnetic resonance (CMR), nuclear imaging techniques including positron emission tomography (PET) and single-photon emission computed tomography (SPECT), and cardiac computed tomography (CT). Recent advances in imaging biomarkers, parametric mapping, myocardial strain analysis, and emerging technologies such as artificial intelligence (AI), radiomics, and molecular imaging were also considered. Multimodality cardiovascular imaging plays a central role in the early detection and comprehensive assessment of myocardial involvement in SSc. Advanced imaging techniques enable improved identification of subclinical myocardial dysfunction, microvascular impairment, inflammation, and fibrosis. An integrated imaging approach combining echocardiography, CMR, nuclear imaging, and CT may facilitate earlier diagnosis, enhance risk stratification, and ultimately improve cardiovascular outcomes in patients with SSc. Full article

Background: Artificial intelligence (AI) and deep learning (DL) are rapidly changing the field of diagnostics and imaging in cardiology, offering tools for automatic segmentation, quantification of changes, and risk stratification. These technologies have the potential to increase diagnostic accuracy, work efficiency, and individualization of patient care. Methods: This structured narrative review critically evaluates clinically validated applications of artificial intelligence (AI) and deep learning (DL) in cardiovascular medicine, focusing on imaging (echocardiography, coronary CT angiography, cardiac MRI, and ECG), risk stratification, and biomarker integration. A systematic literature search was conducted in PubMed for studies published between January 2015 and December 2026, supplemented by references from key articles. Original English-language studies reporting quantitative clinical outcomes were included, with 78 studies ultimately analyzed. Results: AI and DL models, including convolutional neural networks and transformers, achieved performance comparable to experts in cardiac imaging, myocardial perfusion assessment, valve defect detection, and coronary event prediction. Multimodal approaches improved diagnostic accuracy and reproducibility, while explainable AI enhanced transparency and clinical confidence. Deep learning also enabled faster image acquisition and processing without compromising precision. Conclusions: AI and DL have transformative potential in cardiology, offering fast, accurate, and scalable diagnostic tools. The integration of multimodal data, the validation of algorithms in prospective studies, and ensuring the transparency of models are key. Future research should focus on prospective, multicenter validations and the ethical and safe implementation of AI in everyday clinical practice. Full article

Accurate prediction of the state of health (SOH) of lithium-ion batteries is essential for ensuring the safety and reliability of electric vehicles and energy storage systems. While machine learning (ML)-based models have demonstrated strong predictive performance, their limited interpretability remains a major challenge for deployment in safety-critical applications. Although explainable boosting machines (EBMs) provide an interpretable alternative through their additive structure, existing studies still rely on manual analysis of model outputs, which restricts scalability and reproducibility. To address this limitation, this study proposes a structured interpretation framework that integrates EBMs with multimodal large language models (MLLMs). The proposed framework employs EBMs to generate SOH predictions along with global feature importance and variable-level score-density visualizations. These outputs are subsequently processed by an MLLM to perform automated interpretation at both global and variable levels, followed by aggregation, cross-validation, and generation of a unified interpretation report. Experiments were conducted on a lithium-ion battery degradation dataset and the EBM achieved competitive predictive performance compared to baseline ML models. In addition, the quality of the generated interpretations was evaluated using both an MLLM-as-a-Judge and a user study. The evaluation results show that the generated interpretations consistently achieved high scores, with average ratings exceeding 4.5 out of 5 across key criteria such as interpretation accuracy and faithfulness, as assessed by both independent MLLMs and domain experts. These results demonstrate that the proposed framework enables reliable and scalable interpretation of battery SOH prediction models, providing a practical solution for explainable artificial intelligence in battery health management. Full article

The demand for advanced detection methods and accurate staging remains a global challenge in cancer diagnosis. Even though traditional deep learning models in medical imaging achieve high precision, they suffer from limited explainability and multimodal reasoning due to their black-box nature, thereby limiting their clinical applicability. To address this gap, recent research has increasingly explored multimodal approaches that integrate visual and textual clinical data to enhance diagnostic accuracy and interpretability. This study presents a bibliometric analysis of 408 publications from 2021 to 2025, collected from Web of Science and Scopus, using VOSviewer and R-Bibliometrix to map citation networks, co-authorship, and keyword co-occurrences. The results reveal a rapid growth from 1 publication in 2021 to 269 in 2025, with significant contributions from leading countries and institutions. Thematic analysis indicates a shift from conventional convolutional approaches toward transformer-based and self-supervised methods, alongside increasing attention to multimodal learning in cancer imaging tasks such as breast, lung, and brain cancer analysis. Overall, this study provides a structured overview of the evolving research landscape, highlighting key trends, emerging themes, and research gaps to inform future developments in multimodal artificial intelligence for cancer diagnosis. Full article

Biochemical detection is fundamental to various scientific disciplines, yet conventional methods still face inherent bottlenecks in achieving rapid, ultrasensitive, and simultaneous multi-target analysis. Terahertz (THz) waves, characterized by their unique spectral fingerprinting capabilities and non-destructive properties, have emerged as a compelling platform for advanced biochemical sensing. This review outlines the evolution of THz biochemical sensing over the past two decades, tracing its progression from passive identification toward intelligent perception. We structure this technological trajectory around four core themes: sensitivity enhancement, specific recognition, multi-target visualization, and system intelligence. We first evaluate the fundamental limitations of direct detection techniques, such as THz time-domain spectroscopy (THz-TDS). Building on this, we examine how metamaterial-assisted architectures utilize high-quality-factor resonances to achieve trace-level detection, pushing the limits of detection (LOD) down to the ng/mL or even pg/mL scale, and how surface chemical functionalization provides a molecular lock mechanism for selective targeting in complex samples. Furthermore, we highlight the paradigm shift from single-point spectral measurements to spatially resolved multi-target imaging using pixelated metasurfaces. Finally, the review addresses emerging directions, including dynamically tunable intelligent metasurfaces, multimodal on-chip integration platforms, and the growing integration of artificial intelligence (AI) in inverse design and data interpretation, which achieves classification accuracies exceeding 95% even in complex matrices. By synthesizing these developments, this review provides a comprehensive perspective on the future trajectory of THz sensing technologies. Full article

The electrocardiogram (ECG) is a central tool in cardiovascular diagnostics, yet interpretation requires expertise and remains subject to variability. Multimodal large language models (MLLMs) have shown emerging capabilities in medical image analysis, but their performance in ECG interpretation remains insufficiently characterized. This study evaluated the diagnostic accuracy and inter-run reliability of five MLLMs across ECG interpretation tasks. Thirteen standard 12-lead ECGs were presented to five models (ChatGPT-5.3, Gemini 3.1 Pro, Claude Opus 4.6, Grok 4.1, and ERNIE 5.0) across five independent runs per case, yielding 2275 task-level assessments. Six categorical interpretation tasks (rhythm, electrical axis, PR/P-wave morphology, QRS duration, ST/T-wave morphology, and QTc interval) were compared with expert-consensus ground truth, while heart rate estimation was evaluated using mean absolute error (MAE). Overall categorical accuracy ranged from 52.3% to 64.9%. QRS duration classification achieved the highest accuracy (66.2–90.8%), whereas ST/T-wave assessment showed the lowest performance (20.0–41.5%). Heart rate MAE ranged from 14.8 to 46.7 bpm. A dissociation between diagnostic accuracy and inter-run reliability was observed across models. These findings indicate that current MLLMs do not achieve clinically reliable ECG interpretation performance and highlight the importance of assessing diagnostic accuracy and inter-run reliability when evaluating artificial intelligence systems in biomedical diagnostics. Full article

Cardiac amyloidosis (CA) remains underrecognized due to overlapping features with other cardiovascular conditions, including hypertrophic cardiomyopathy and hypertensive heart disease. Certain ‘red flag’ features across the clinical and imaging spectrum help identify CA. However, these features are often absent, subtle, or inconsistently recognized, particularly in early disease, and are atypical phenotypes. This leads to frequent delays in diagnosis and presentation at advanced stages. Artificial intelligence (AI) offers a promising approach to detect subtle disease signatures by integrating multimodal and longitudinal data beyond human pattern recognition. AI-enhanced electrocardiography has emerged as a scalable screening tool, demonstrating high diagnostic performance and enabling earlier detection. In parallel, echocardiographic AI has evolved toward video-based analysis, improving standardization and reducing inter-reader variability. Similarly, AI applications in cardiac magnetic resonance and nuclear scintigraphy allow for automated quantification and more reproducible assessment of amyloid burden. Beyond diagnosis, emerging models support disease phenotyping, risk stratification, and treatment monitoring. This review synthesizes current applications of AI across multimodal testing in the evaluation and diagnosis of CA. Full article

Breast cancer remains the most frequently diagnosed malignancy in women worldwide, accounting for approximately 2.3 million new cases and 670,000 deaths annually. The diagnostic landscape has undergone a paradigm shift over the past two decades, evolving from morphology-based classification toward molecularly informed, precision-guided strategies. Early and accurate diagnosis is fundamental to improving outcomes; advances in imaging technology, including digital breast tomosynthesis (DBT), contrast-enhanced mammography (CEM), and abbreviated magnetic resonance imaging (MRI), have improved sensitivity and specificity in diverse patient populations. Simultaneously, the integration of artificial intelligence (AI) and radiomics into screening workflows offers unprecedented potential for risk stratification and a reduction in false-positives. At the pathological level, multi-gene expression profiling assays such as Oncotype DX, MammaPrint, Prosigna, and EndoPredict have refined prognostic classification and guide adjuvant chemotherapy decisions in early-stage hormone receptor-positive disease. The emergence of liquid biopsy, circulating tumor DNA (ctDNA), circulating tumor cells (CTCs), and exosomal biomarkers provides minimally invasive tools for real-time monitoring of response, residual disease, and the evolution of resistance mechanisms. Precision diagnostics now encompass next-generation sequencing (NGS)-based comprehensive genomic profiling, enabling identification of actionable alterations such as PIK3CA mutations, HER2 amplification, BRCA1/2 pathogenic variants, and NTRK fusions, each linked to approved therapeutic agents. The purpose of this review is to provide a comprehensive synthesis of current and emerging diagnostic modalities in breast cancer—from population-level screening to individualized molecular profiling—and to examine how integrative, multimodal diagnostic platforms are reshaping clinical decision-making in the era of precision medicine. Full article

Large Language Models (LLMs) have become foundational to modern Artificial Intelligence (AI), enabling advanced reasoning, multimodal understanding, and scalable human-AI interaction across diverse domains. This survey provides a comprehensive review of major proprietary and open-source LLM families, including GPT, LLaMA 2, Gemini, Claude, DeepSeek, Falcon, and Qwen. It systematically examines architectural advancements such as transformer refinements, mixture-of-experts paradigms, attention optimization, long-context modeling, and multimodal integration. The paper further analyzes alignment and safety mechanisms, encompassing instruction tuning, reinforcement learning from human feedback, and constitutional frameworks, and discusses their implications for controllability, reliability, and responsible deployment. Comparative analysis of training strategies, data curation practices, efficiency optimizations, and application settings highlights key trade-offs among scalability, performance, interpretability, and ethical considerations. Beyond synthesis, the survey introduces a structured taxonomy and a feature-driven comparative study of over 50 reconstructed LLM architectures, complemented by an interactive visualization interface and an open-source implementation to support transparency and reproducibility. Finally, it outlines open challenges and future research directions related to transparency, computational cost, data governance, and societal impact, offering a unified reference for researchers and practitioners developing large-scale AI systems. Full article

Botulinum neurotoxin (BoNT) treatment outcomes are commonly assessed through visual evaluation of facial wrinkle patterns, a process that remains inherently subjective despite structured grading systems. This study evaluated whether contemporary multimodal artificial intelligence (AI) systems can identify facial changes associated with BoNT treatment, using region-specific wrinkle patterns as surrogate markers of underlying muscle activity. A dataset of 46 facial images (23 pre-treatment, 23 post-treatment) was analyzed using four multimodal models, each assessed across five independent runs. Models were tasked with classifying treatment state from single images, detecting wrinkle presence in the forehead, glabella, and periorbital regions, and generating exploratory severity scores and age estimates. Two models achieved 100% accuracy in distinguishing pre- from post-treatment images in this dataset, while region-specific wrinkle detection was variable and frequently did not exceed majority-class baselines. Inter-run reliability varied substantially across models. Exploratory wrinkle severity scores showed directional differences between treatment states, whereas apparent age estimates demonstrated minimal systematic variation. These findings suggest that global facial changes associated with BoNT treatment appear to be detectable in model outputs, but region-specific assessment remains limited, underscoring the need for cautious interpretation and further validation. Full article

Aerospace vehicles operate across a wide flight envelope, traversing dense atmospheric layers from near-space to low Earth orbit. Trajectory planning and optimization in a large spatial domain and wide speed range present severe challenges to traditional methods, including computational efficiency, model accuracy, and constraint adaptability. Artificial intelligence provides an effective pathway to overcome these limitations and has become a key driver for advancing trajectory planning and optimization of aerospace vehicles. This paper presents a systematic review of the core characteristics of aerospace trajectory planning, including environment coupling, multi-constraint compliance, propulsion integration, and aerodynamic nonlinearity, as well as the limitations of traditional methods. The study focuses on the application of intelligent algorithms in both the ascent and reentry phases. For the ascent phase, three key issues are addressed: multistage hybrid optimization with continuous and discrete variables, propulsion multimodal–trajectory coupling, and trajectory reconfiguration under engine failure. For the reentry phase, discussions are focused on such technical difficulties as multi-constraint trajectory generation, no-fly zone avoidance, and multi-mission requirement optimization. Finally, future research directions in intelligent trajectory planning and optimization are discussed, providing theoretical support and methodological guidance for the autonomous and intelligent development of aerospace vehicle trajectory planning. Full article

In highly digitalized and networked modern trading systems, large volumes of heterogeneous data are continuously generated from multiple sources during market operations. However, due to the complexity of data structures, significant differences in temporal scales, and constraints imposed by data privacy protection, traditional single-source modeling approaches are unable to fully exploit multisource information. To address this issue, a federated multimodal prediction framework for complex market systems, termed Federated Market-Sensor Transformer (FMST), is proposed. In this framework, data originating from different information sources are uniformly modeled as multimodal time series. A multimodal market-sensor representation module is constructed to perform unified feature encoding, and a cross-modal Transformer fusion architecture is employed to characterize dynamic interaction relationships among different information sources. Meanwhile, a federated collaborative learning mechanism is introduced during the training phase, enabling multiple data nodes to perform collaborative model optimization without sharing raw data. In this manner, data privacy can be preserved while improving the cross-region generalization capability of the model. Systematic experimental evaluation is conducted on the constructed multimodal market-sensor dataset. The experimental results demonstrate that the proposed method consistently outperforms traditional statistical models and deep learning approaches across multiple evaluation metrics. In the main prediction experiment, FMST achieves a root mean square error (RMSE) of $0.1136$, a mean absolute error (MAE) of $0.0832$, and a coefficient of determination $R^2$ of $0.8517$, while the direction prediction accuracy reaches $74.56\%$, clearly outperforming baseline models including ARIMA, LSTM, Temporal CNN, Transformer, and FedAvg-LSTM. In the cross-region generalization experiment, FMST maintains strong performance, achieving an RMSE of $0.1242$, an MAE of $0.0908$, an $R^2$ value of $0.8261$, and a direction prediction accuracy of $72.48\%$. The ablation study further indicates that the three core components—multimodal market-sensor representation, cross-modal Transformer fusion, and federated collaborative learning—each make important contributions to the overall model performance. These experimental findings demonstrate that the proposed method can effectively integrate multisource market information and significantly enhance the prediction capability for complex market dynamics, providing a new technical pathway for the application of artificial intelligence-driven multimodal sensing systems in economic data analysis. Full article

Personalized transdermal drug delivery systems (TDDS) represent a transformative approach in precision medicine by enabling patient-specific, non-invasive, and controlled therapeutic administration. Conventional transdermal patches are limited by fixed dosing, passive diffusion, and interindividual variability in skin permeability and metabolism, often leading to suboptimal therapeutic outcomes. Recent advances in materials science, nanotechnology, microneedle engineering, and digital health have enabled the development of next-generation personalized TDDS capable of programmable, adaptive, and feedback-controlled drug release. Smart wearable patches integrating biosensors, microfluidics, microneedles, and wireless connectivity allow real-time monitoring of physiological and biochemical parameters, enabling closed-loop drug delivery tailored to individual metabolic profiles. Nanocarriers such as lipid nanoparticles, polymeric nanoparticles, and stimuli-responsive hydrogels further enhance drug stability, penetration, and controlled release, while 3D-printing technologies facilitate patient-specific customization of patch geometry, drug loading, and release kinetics. Artificial intelligence (AI) and machine learning tools are increasingly being employed to predict drug permeation behavior, optimize enhancer combinations, and personalize dosing regimens based on pharmacogenomic and pharmacokinetic data. Despite these advances, regulatory complexity, manufacturing standardization, long-term biocompatibility, and cybersecurity considerations remain critical challenges for clinical translation. This review highlights recent innovations in personalized TDDS, discusses their clinical potential, and examines regulatory and technological barriers. Collectively, these emerging smart transdermal platforms offer a promising pathway toward adaptive, patient-centered therapeutics that can significantly improve treatment efficacy, safety, and compliance. Future research should focus on integrating multimodal biosensing, advanced biomaterials, scalable manufacturing strategies, and robust regulatory frameworks to enable clinically validated, fully autonomous transdermal systems that can dynamically adapt to real-time patient needs in diverse therapeutic settings. Full article

Background/Objectives: Multimodal large language models (MLLMs) are increasingly evaluated for diagnostic tasks in medical imaging, including radiographic interpretation. However, most studies focus primarily on binary fracture detection and rarely assess clinically relevant fracture characteristics such as displacement or intra-articular extension, which influence treatment decisions. In addition, most evaluations rely on single-run inference designs that do not assess response reproducibility. This study evaluated the diagnostic performance and inter-run reliability of five MLLMs for radiographic assessment of distal radius fractures. Methods: Fifty fracture-positive distal radius radiographs were evaluated by five MLLMs (ChatGPT 5.3, Gemini 3.1 Pro, Claude Opus 4.6, Grok 4.1, and ERNIE 5.0) across five independent zero-shot inference runs (n = 1250 observations). Diagnostic tasks included fracture detection, intra-articular extension, and displacement. Sex and age were exploratory endpoints. Performance was summarized using sensitivity (fracture detection) and accuracy (other tasks), with inter-run reliability assessed via Fleiss’ κ. Results: Performance varied across tasks and models. Fracture detection sensitivity ranged from 39.6% to 99.6%, with two models exceeding 90%. Intra-articular extension accuracy ranged from 51.6% to 55.6%, consistent with chance-level performance. Displacement classification ranged from 34.8% to 70.4%. One model achieved substantial inter-run agreement across binary tasks (κ > 0.60), whereas two models showed slight agreement (κ < 0.20). Conclusions: Only two models exceeded 90% sensitivity for fracture detection, while intra-articular extension remained at chance level (≤55.6%). Substantial inter-run reliability (κ > 0.60) was observed in only one model. These findings indicate that current MLLMs do not reliably support multidimensional fracture assessment and that single-run evaluations overestimate robustness. Full article

Chronic kidney disease is rising worldwide, and kidney transplantation remains the preferred modality of kidney replacement therapy. However, long-term graft survival continues to be limited by chronic alloimmune injury, particularly antibody-mediated rejection (ABMR) and its chronic active form. This narrative review synthesizes contemporary evidence on the immunopathogenesis, epidemiology, diagnosis, and management of kidney allograft rejection, with a deliberate focus on studies from the last five years and on United States and European cohorts. We summarize current concepts of T cell–mediated rejection (TCMR), ABMR, mixed and donor-specific antibody (DSA)–negative phenotypes, and the evolution of the Banff classification, highlighting how chronic active ABMR has emerged as a leading cause of death-censored graft loss. We then critically appraise the conventional diagnostic triad of creatinine/eGFR, DSA, and biopsy and review emerging tools, including donor-derived cell-free DNA, urinary chemokines such as CXCL9 and CXCL10, additional blood- and urine-based biomarkers, and biopsy transcriptomics. We also examine how artificial intelligence and machine learning may support digital pathology, multimodal risk prediction, and data integration, while recognizing the current challenges of biological interpretability, external validation, and clinical implementation. Finally, we propose a rejection-focused precision-medicine framework and outline key research gaps, including multicenter validation, trial-ready endpoints, and governance for AI-enabled pathways. Overall, the field is moving from isolated diagnostic signals toward integrated, biologically informed, and clinically actionable approaches to rejection detection and risk stratification. Full article