搜索结果：view

Mammography screening is an essential tool for early detection of breast cancer. The speed and accuracy of mammography interpretation has the potential to be improved with deep learning methods. However, the development of a foundation visual language model (VLM) is hindered by limited data and domain differences between natural and medical images. Existing mammography VLMs, adapted from natural images, often ignore domain-specific characteristics, such as multi-view relationships in mammography. Unlike radiologists who analyze both views together to process ipsilateral correspondence, current methods treat them as independent images or do not properly model the multi-view correspondence learning, losing critical geometric context and resulting in suboptimal prediction. We propose GLAM: Global and Local Alignment for Multi-view mammography for VLM pretraining using geometry guidance. By leveraging the prior knowledge about the multi-view imaging process of mammograms, our model learns local cross-view alignments and fine-grained local features through joint global and local, visual-visual, and visual-language contrastive learning. Pretrained on EMBED [14], one of the largest open mammography datasets, our model outperforms baselines across multiple datasets under different settings.

Photoacoustic computed tomography (PACT) combines the high optical absorption contrast of optical excitation with the deep tissue penetration enabled by ultrasonic detection, making it a promising imaging modality. However, constraints on transducer density and angular coverage often result in sparse-view acquisitions that cause severe artifacts. In this work, we propose ND-net, a progressive dual-branch adversarial diffusion framework for efficient and high-quality sparse-view PACT reconstruction. The framework uses two stages where a residual artifact-reconstruction branch estimates structured sparse-view artifacts, followed by an adversarially guided full-view diffusion branch that refines structural information. By enabling flexible reverse transitions, ND-net supports large-step diffusion sampling with only four reverse iterations, improving inference efficiency. Experiments on simulated vessel data, circular phantom measurements, and in vivo mouse abdomen imaging demonstrate improved reconstruction quality over representative analytical and learning-based methods under highly sparse acquisition conditions. These results indicate that ND-net improves sparse-view PACT reconstruction while enabling efficient inference.

Cognitive function is closely linked to brain energy metabolism and may be compromised by aging, metabolic stress, and neuropsychiatric disease. Ketone bodies can serve as an alternative cerebral fuel and may also exert signaling effects relevant to cognition. Exogenous ketones (EK) offer a practical means of increasing circulating ketone concentrations without dietary carbohydrate restriction. However, the overall effect of EK supplementation on cognitive performance in humans has not been systematically quantified. A systematic review and meta-analysis were conducted in accordance with PRISMA guidelines. PubMed, Web of Science, and Embase were searched through October 2025 for randomized controlled trials investigating the effects of EK on cognitive outcomes in healthy adults or individuals with neuropsychiatric conditions. Data extraction and quality assessment were performed independently by multiple reviewers using the PEDro scale. Standardized mean differences (SMD) were calculated using random-effects models. Subgroup and meta-regression analyses examined the influence of ketone formulation, intervention duration, dose, population type, and presence of acute cognitive stressors. 38 studies comprising 41 protocols (1,602 participants) were included in the systematic review, with 29 protocols (1,117 participants) eligible for meta-analysis. EK supplementation was associated with a statistically significant improvement in cognitive performance compared with placebo (SMD&#x202f;=&#x202f;0.29, 95% CI 0.16-0.41; p&#x202f;<&#x202f;0.001). Sub-group analyses did not show statistically significant differences between the type of supplementation (p&#x202f;=&#x202f;0.083), study duration (acute vs. intermediate; p&#x202f;=&#x202f;0.11), population type (healthy vs. Alzheimer's disease; p&#x202f;=&#x202f;0.077), or the presence of acute cognitive stressors (p&#x202f;=&#x202f;0.89). Meta-regression revealed a positive association between daily EK dose and cognitive improvement. EK supplementation is associated with modest improvements in cognitive performance across diverse populations and study designs. These findings support EK as a flexible nutritional strategy for cognitive support and warrant further investigation in well-powered, long-term trials to clarify optimal dosing, formulation, and clinical applicability. https://www.crd.york.ac.uk/PROSPERO/view/CRD42023471727, CRD42023471727.

As surgical AI transitions from pixel-level detection to complex reasoning, Scene Graphs (SGs) offer the structured, relational representations necessary to decode dynamic surgical environments. This PRISMA-ScR-guided scoping review systematically maps the evolving landscape of SG research in surgery, analyzing 52 primary studies to chart applications and methodological shifts. Our analysis reveals rapid growth, yet uncovers a critical 'data divide': internal-view research (e.g., triplet recognition from endoscopic video) accounts for 79% of studies and predominantly uses real-world 2D video, while external-view operating room modeling relies heavily on simulated data. Methodologically, we identify a decisive shift from foundational graph neural networks to specialized foundation models and generative AI, which together now account for approximately 50% of research in 2025. Crucially, our synthesis suggests that Scene Graphs are evolving from simple descriptors into essential 'neuro-symbolic guardrails', providing the structured, verifiable intermediate representation needed to prevent hallucinations in increasingly autonomous Surgical Foundation Models. Despite this promise, a major translational gap remains: none (0/52) of the reviewed studies have proceeded to prospective clinical validation. We conclude that bridging this gap requires moving beyond standard computer vision metrics; we therefore propose the 'Validation Trinity' - prioritizing Semantic Query Success, Latency-Aware Accuracy, and Safety-Critical Recall - as the necessary evaluation framework to bring graph-based surgical AI into clinical practice.

Mental health difficulties are highly prevalent worldwide. Digital mental health tools (DMHTs) have been developed to increase accessibility to mental healthcare for people who may struggle to access care due to cost, location or stigma. As the views of stakeholders are important in understanding the potential barriers to and facilitators of DMHT implementation, the aims of this review were to critically appraise and synthesise qualitative findings relating to the perceptions and/or experiences of healthcare professionals (HCPs) on the use of digital mental health tools in clinical practice. A systematic search of mixed-method and qualitative studies was performed using five databases. Eligible studies were quality-assessed. Data were analysed using inductive thematic synthesis. Fifteen studies were identified and reviewed. Four main themes (alongside eight subthemes) were developed from the data of 604 HCPs: 1) DMHTs should augment - not replace - face-to-face clinical care; 2) Considerations and caveats to use in clinical practice; 3) Using DMHTs to enhance clinical care; and 4) Perceived barriers and concerns. HCPs strongly endorsed the view that DMHTs offer increased access to care, however, concerns about their therapeutic quality, risk management, and workload burden persist. Context-sensitive implementation and proper infrastructure are essential for successful integration into mental health services. https://www.crd.york.ac.uk/prospero/, identifier CRD42020188879.

Various comorbidities have been associated with psoriasis. Most clinical studies support the hypothesis that psoriasis may be a risk factor for dementia. Meanwhile, some evidence indicates that certain immunomodulatory agents, many of which are widely used in psoriatic disease management, exert neuroprotective effects and may attenuate dementia progression. In view of the lack of existing studies that specifically investigate the effects of systemic treatments for psoriatic disease on dementia or cognitive impairment, in this narrative review, we focus on Alzheimer's disease, as a model to explore whether systemic psoriasis treatments influence dementia risk and severity. Our findings suggest that some systemic treatments for psoriasis may also provide potential neuroprotective benefits.

Currently, numerous studies have employed machine learning (ML) methods to develop predictive models for depression risk in patients with diabetes mellitus (DM); however, the findings remain inconsistent. Therefore, this study aims to clarify the current state of research and emerging trends in this field by systematically evaluating the performance, strengths, and limitations of existing prediction models. This systematic review evaluates the performance and clinical applicability of ML-based depression risk prediction models for patients with DM, providing reliable evidence to assist healthcare professionals in selecting and optimizing more appropriate prediction models. We conducted a systematic search of clinical studies employing ML approaches to predict depression risk in patients with DM across the PubMed, Embase, Cochrane Library, and Web of Science databases, from their inception to January 2026. The primary performance metric for the models was the area under the receiver operating characteristic curve (AUC) along with its 95% confidence interval (95% CI). Two independent researchers screened the literature, extracted data, and used PROBAST-AI to assess the risk of bias and clinical applicability of the included studies. Pooled AUC was estimated using the Der Simonian and Laird random-effects model. A total of 14 studies comprising 64 distinct ML models were included. All included studies were assessed as high risk of bias and high clinical applicability. A pooled analysis of the best-performing ML prediction models reported in each study showed a pooled AUC of 0.822 (95% CI, 0.789-0.858), indicating relatively good overall predictive performance. However, there was substantial heterogeneity among the studies (I&#xb2; =&#xa0;97.4%; P <&#xa0;0.001). Subgroup analysis based on ML model types revealed the following pooled AUC values: 0.765 (95% CI 0.706-0.829) for traditional regression models, 0.789 (95% CI 0.747-0.834) for general machine learning models, and 0.802 (95% CI 0.769-0.836) for deep learning models. Notably, logistic regression (LR) (n = 10) was the most frequently employed ML method for developing depression risk prediction models in patients with DM. To evaluate model generalizability and avoid overfitting, the included studies adopted three validation strategies: 5-fold cross-validation yielded a pooled AUC of 0.913 (95% CI 0.781-1.067), 10-fold cross-validation yielded 0.819 (95% CI 0.781-0.858), and random split validation yielded 0.747 (95% CI 0.648-0.862). The most commonly used predictors in the included models were age, sex, and body mass index (BMI), which are readily available in clinical settings and strongly associated with depression risk. ML-based depression risk prediction models for patients with DM demonstrate overall satisfactory predictive performance. However, most existing studies had relatively small sample sizes and lacked external validation. Future research should prioritize refining study design and optimizing clinical data processing to improve the generalizability and stability of these models in clinical practice. https://www.crd.york.ac.uk/PROSPERO/view/CRD420251243343, identifier CRD420251243343.

Heart failure (HF) is a global medical condition marked by substantial morbidity, mortality, and healthcare costs with complex pathophysiology and variation in definitions. Machine learning (ML) has emerged as a promising approach to improve HF classification and risk prediction by leveraging various data sources. This study aims to present the current state-of-the-art multimodal ML models for HF classification and prognosis prediction, focusing on their modalities, performance, and clinical utility. Following PRISMA guidelines and registered with PROSPERO (CRD420250654631), this review searched across four electronic databases (November 2014 - November 2024) and identified 284 unique records, of which 15 were included in the final synthesis. The quality of the studies was evaluated using QUADAS-2 and QUAPAS. Our results showed that the two most common multimodal combinations were tabular-image and tabular-text. The algorithms of the models included convolutional neural networks for image data, transformer-based approaches for text, with well-known fused techniques (early, middle, late fusion). Overall, multimodal models demonstrated superior performance compared to unimodal approaches, achieving area under the receiver operating characteristic curve values frequently exceeding 80% and reaching as high as 98.2%. Despite promising results, challenges include inconsistent reporting of performance metrics and their 95% confidence intervals, limited external validation, a near absence of prospective studies, and a deficiency in integrating genetic or 'omics' information with conventional data. These challenges must be addressed to promote clinical adoption and future research. https://www.crd.york.ac.uk/PROSPERO/view/CRD420250654631, identifier CRD420250654631.

Post-translational modifications (PTMs) are crucial regulatory mechanisms that modulate the structure, function, and stability of proteins, playing an essential role in the regulation of cellular processes. Dysregulated PTMs are associated with various aspects of cancer development, including uncontrolled cell growth, evasion of apoptosis, metastasis, and drug resistance. This review offers a detailed examination of several major PTMs, including phosphorylation, acetylation, ubiquitination, SUMOylation, and methylation, discussing their distinct roles in cancer biology. It also provides an in-depth analysis of the latest advancements in the study of PTMs in cancer biology, focusing on the mechanisms by which these modifications contribute to tumorigenesis and their potential as therapeutic targets. It highlights the significant progress made in the identification of PTMs across different cancer types, emphasizing the role of PTMs in shaping cancer progression and immune modulation. Additionally, the paper discusses cutting-edge technologies, particularly mass spectrometry and computational proteomics, that have revolutionized the detection and characterization of PTMs. These advancements have enabled the identification of novel cancer biomarkers and therapeutic targets, offering new avenues for early detection, prognostic monitoring, and the development of targeted therapies in cancer treatment.

Urban disaster simulation plays a crucial role in facilitating proactive mitigation strategies and enhances urban resilience. However, intuitive and precise methods for visualizing and pre-enacting prospective disaster scenarios have not received sufficient attention. This study proposes a novel framework that leverages generative artificial intelligence (AI) to pre-enact and evaluate urban scenarios under multiple extreme weather events across 20 climate-vulnerable cities in China. The framework integrates multimodal generative AI models with a large dataset comprising 1620 disaster photographs (2010-2024) and 528 street-view images. Through a workflow that includes disaster scene generation, semantic recognition of street elements, resilience risk assessment, and environmental intervention strategy generation, the proposed approach derives precisely targeted climate adaptation strategies. The semantic similarity between generated disaster scenes and real-world scenes exceeds 0.82. The results indicate that increasing green coverage, improving sidewalk accessibility, enhancing traffic management, and upgrading public service facilities are key measures for improving multi-hazard resilience. Furthermore, the study proposes differentiated resilience strategy packages tailored to three types of disaster scenarios: weather response-oriented, temperature response-oriented, and geological response-oriented scenarios. These findings demonstrate the potential of generative AI tools to support evidence-based urban resilience policy formulation.

暂无摘要（点击查看详情）

The comparative effectiveness of chlorhexidine gluconate (CHG) versus povidone-iodine (PVI) for preventing surgical site infections (SSIs) remains unclear across surgical types and resource settings. This study compared CHG and PVI overall and within key clinical subgroups. Five databases were searched through February 2025 for randomized controlled trials comparing CHG with PVI and reporting SSI outcomes. Random-effects models generated pooled odds ratios (ORs) with 95% confidence intervals (CIs). Twenty-nine RCTs involving 35,317 patients were included. CHG significantly reduced superficial incisional SSIs (OR = 0.80; 95% CI 0.67-0.95; p = 0.01; I2 = 18.4%), but not overall, deep, or organ/space infections; meta-regression indicated that patient age was a significant effect modifier. In cesarean sections, CHG lowered overall (OR = 0.64; 95% CI 0.48-0.85), superficial (OR = 0.65; 95% CI 0.48-0.87), and deep incisional SSIs (OR = 0.41; 95% CI 0.22-0.75). In abdominal surgery, CHG reduced only superficial incisional SSIs (OR = 0.68; 95% CI 0.52-0.91). No significant differences were observed in gynecologic, cardiothoracic, or orthopedic procedures. By wound classification, CHG had no effect in clean surgery but reduced superficial incisional SSIs in clean-contaminated cases (OR = 0.65; 95% CI 0.48-0.89). By income level, no differences were seen in high-income countries, while in low- and middle-income countries CHG decreased overall (OR = 0.58; 95% CI 0.46-0.74), superficial (OR = 0.54; 95% CI 0.38-0.76), and deep incisional SSIs (OR = 0.48; 95% CI 0.25-0.92). Alcohol-based CHG and alcohol-based PVI are comparably effective in most surgical settings. However, CHG demonstrates superior prevention of SSIs in cesarean, abdominal, and clean-contaminated surgeries, with the most substantial benefit in low- and middle-income settings. Broader use may be justified pending cost-effectiveness evaluation.

暂无摘要（点击查看详情）