搜索结果：create

This study aimed to create machine learning algorithms using the Australian & New Zealand Society of Cardiac & Thoracic Surgeons (ANZSCTS) Database that can predict readmissions or post-discharge mortality within 30 days of cardiac surgery. Data from 54 public and private Australian hospitals from January 2017 to December 2021 were included in this study. Only coronary artery bypass grafting (CABG), valvular heart surgery, or aortic surgery cases were included for analysis. The primary outcome of this study was the performance of our machine learning models measured using sensitivity, specificity, positive and negative predictive values, accuracy, area under the receiver operating characteristic curve (AUROC), and area under the precision-recall curve. Of the 178,252 patients in the database, 61,721 CABG, valvular heart surgery, and aortic surgery cases were identified. The incidence of 30-day readmission or post-discharge mortality was 10.3% (6,351/61,721). The best-performing model was the deep neural network (AUROC, 0.615). A calibrated version of this model predicted one in five patients at higher risk of readmissions or post-discharge mortality compared with the lowest risk group (2.4× increase in incidence rate and an odds ratio of 2.7 [2.2-3.2]). Using machine learning, models developed from the Australian ANZSCTS Database demonstrated limited predictive ability for readmissions or post-discharge mortality. The ANZSCTS Database may require additional informative variables before models become sufficiently accurate. Individual hospitals or small hospital networks may be best placed to incorporate routinely collected information (such as electronic medical records) alongside ANZSCTS data to create more accurate models.

Advances in the realism of synthetic media created with generative adversarial networks (GANs), diffusion models, and face manipulation tools has created an increased demand for well-established deepfake detection systems that can detect many different types of manipulation artifacts. However, most single model deepfake detectors are not very robust because they rely on specific forensic cues and do not adapt well to shifts in how synthesis occurs. We present DeepFakeBuster as a confidence-calibrated adaptive ensemble for deepfake image detection by fusing together heterogeneous deep learning models built around detecting complementary forensic cues e.g., spatial inconsistencies, boundary artifacts, noise residuals, semantic consistency, and frequency-domain features. In contrast to traditional ensemble approaches that use static averaging of detector outputs, our proposed framework utilizes reliability aware adaptive fusion where the contribution of each detector to the fused output is adjusted dynamically through the use of reliability priors derived from validation and input-specific confidence estimates. Our experimental evaluation on a dataset comprised of 192,000 authentic and manipulated images shows that our ensemble significantly outperforms both individual constituent detectors as well as static fusion baselines, with an overall accuracy of 97.8% for the evaluated conditions. Additionally, an interpretable forensic analysis module provides visual and quantitative indicators associated with manipulation-sensitive regions. The findings suggest that confidence-aware heterogeneous ensemble learning represents a promising direction for robust deepfake detection.

The tumour microenvironment imposes severe metabolic constraints that reshape anti-tumour immunity across the cancer-immunity cycle. Rather than serving merely as passive byproducts of tumour growth, tumour-derived metabolites and nutrient imbalances act as potent metabolic checkpoints-stage-specific barriers that disrupt the functional progression of dendritic cells (DCs) and T cells from antigen presentation to effective tumour clearance. In this review, we propose a framework that overlays the cancer-immunity cycle with major metabolic checkpoints, including glucose and amino acid competition, acidosis and lipid overload, to clarify how distinct metabolic stresses create immune bottlenecks at different stages of the anti-tumour response. We then discuss how distinct tumour metabolic phenotypes, characterized by high glycolysis, amino acid dependency or lipid dysregulation, generate local environmental stresses that differentially reprogram DC function and T cell fitness. Particular emphasis is placed on the DC-T cell axis as a critical site where multiple metabolic defects converge, destabilizing antigen presentation, co-stimulation and immunological synapse function. We further survey emerging therapeutic strategies aimed at restoring the DC-T cell axis and effective anti-tumour immunity, ranging from small-molecule metabolic inhibitors to metabolically engineered adoptive cell therapies designed to function in hostile microenvironments. Finally, we highlight emerging technologies such as single-cell and spatial multi-omics, real-time metabolic imaging and microphysiological systems that can resolve the spatiotemporal heterogeneity of tumour immunometabolism and support more precise immunometabolic interventions.

Standard of care RPT may result in dose-limiting side effects of the salivary glands include sialadenitis and xerostomia that are inconsistent with outcomes from external beam for equivalent ADs. This work develops a parametrizable 'Macro-to-Micro' (M2μ) model that demonstrates the consideration of small-scale dosimetry in comparison to conventional methods that typically assume uniform voxel or organ-level uptake. Anatomical features of the salivary gland are represented by annular structures (a centralized branching network of ducts, including excretory, lobar, interlobular) and small-scale voxels (intralobular ducts and acinar cells) with dimensions set to reproduce ex-vivo murine histopathology measurements and scaled up in size for extrapolation to humans. Simulations were performed by scoring and recording S-value histograms to the target ductal and acinar cells, for two beta emitters (177Lu, 131I) and one alpha emitter (225Ac). Four idealized activity distributions were created to assign activity to surface and volumes of the annuli, and GEANT4 v11 was used for radiation transport calculations. Comparisons against whole gland uniform spherical self-dose S-value calculations were conducted to validate the radiation transport and to highlight differences between M2μ and conventional dosimetry approaches. Analyses for both models showed greater S-value variation in comparison to the homogeneously distributed activity S-value calculation. The most notable result was that the calculated S-values differed between different branches, depending on the geometric size of the annuli. For the surrounding acinar cells, S-values from ductal cells decreased as a function of distance from the branching structures. High variability of S-values, depending on ductal cell dimensions as well as within the acinar cell distribution of the salivary gland highlights the potential clinical utility of small-scale approaches to RPT salivary gland dosimetry, contingent on clinical translation and validation. To this end, future work should further refine the model and incorporate small-scale activity distributions from pre-clinical and translational studies.

Vertebral bodies are anatomical landmarks for localizing measurements in applications such as body composition analysis. This study aimed to provide open-access vertebral body labels and deep-learning segmentation models, and to compare their performance in identifying the third lumbar vertebra (L3) with existing solutions. Thoracic and lumbar vertebral body labels were created for 1460 CT scans from two public datasets. Two residual-encoder nnU-Net models were trained on 1216 cases, and a two-step pipeline was developed. Segmentation performance was tested on 244 cases. In an independent oncological dataset of 300 cases, three readers manually annotated L3 landmarks. Labeling performance was assessed using signed and absolute center localization error (cranio-caudal distance between predicted and annotated L3 center) and center hit rate (predicted center within annotated vertebral body), followed by a comparison with existing models. The two models and the two-step pipeline achieved Dice scores of 0.962 [95% CI, 0.941 to 0.978], 0.939 [95% CI, 0.907 to 0.965], and 0.953 [95% CI, 0.926 to 0.974] for segmentation of the L3 vertebral body. Center hit rates were similarly high across models. Signed center localization error did not differ significantly, whereas absolute center localization error was lower for VertebralBodiesL and VertebralBodiesStepwise than for the vertebral-body-specific TotalSegmentator pipeline. SPINEPS / VERIDAH results are reported separately because vertebral definitions differed. The proposed models accurately segment thoracic and lumbar vertebral bodies and reliably identify L3. Labels, model weights, and a body composition analysis pipeline are openly available.

Community Advisory Boards (CABs), adopted from the HIV response, have become the most institutionalised form of community engagement in tuberculosis (TB) research. However, little work has examined the historical-political processes shaping their development or how these dynamics influence community participation in TB research governance. This qualitative study draws on 28 semi-structured interviews with TB advocates, community engagement practitioners, and TB CABs representatives across countries and levels of global health governance. Using the metaphor of "space" drawn from participatory development, the analysis examines how CABs have emerged, evolved, and operated within TB research over time. CABs were initially mobilised by HIV activists to create spaces within TB research, as broader TB social movements had proven difficult to cultivate. As the CAB model became increasingly institutionalised as a formal engagement mechanism, they were intended to become sites for nurturing TB advocates and enabling their participation in research decision-making. However, sustaining these activist ambitions remained precarious as TB CABs were under-resourced and their knowledge transmission emphasised the technical over the radical, narrowing engagement into more operational and manageable forms. Within these tightly managed spaces, the difficulty of demonstrating the value of CABs reflects how the struggle for power redistribution has shifted from the politics of representation to that of evidence. TB CABs embodies communities' agency and creativity to insert and assert their voice within a biomedical landscape that structures their participation unequally. Reframing CABs as political and relational spaces rather than instrumental mechanisms may strengthen their transformative potential within TB research and policy.

Partners in Contraceptive Choice and Knowledge (PICCK) was a time-limited (2018-2023) initiative to improve access to and quality of contraceptive care at Massachusetts birth hospitals and included continuous involvement of a community advisory board (CAB). In addition to providing strategic guidance, CAB members co-lead didactic activities, helped develop a patient-centered contraceptive decision aid, and created a novel mechanism to provide health literacy and accessibility assessments to clinics. Inclusion of a CAB in PICCK proved feasible and helped ensure equity-focused programming.

Accelerating research advancing health equity for the entire inheritable bleeding disorders community requires a new approach. It must be firmly rooted in health equity, diversity, and inclusion (HEDI) and center the knowledge of people living with inheritable bleeding disorders, the Lived Experience Experts (LEEs). The National Bleeding Disorders Foundation charged seven multidisciplinary working groups (WGs) with developing a National Research Blueprint (NRB) for this Bleeding Disorders Research Collaborative (BDRC). The Infrastructure and Workforce WGs, in collaboration with the HEDI and LEE WGs, met virtually to develop recommendations for BDRC operationalization. A progressive network of elements and processes capacitating diverse community-prioritized research ideas into successfully completed BDRC projects is proposed. Essential components for launch, iterative evolution, effective conduct, and accountability are described. Sharing resources and expertise and embedding research in inheritable bleeding disorders care will create synergistic efficiencies. Education and training to grow and empower interdisciplinary research teams, including LEEs and HEDI champions as valued members, are detailed. Shared leadership integrating LEE and HEDI expertise throughout will provide dynamic governance. BDRC infrastructure and workforce development must start small and grow iteratively in partnership with the many organizations that share its vision of health justice. The National Research Blueprint is a proposal for a new Bleeding Disorders Research Collaborative (BDRC) doing the research people with bleeding disorders need and want. The people who live with a disorder, and their close family members affected by it, are Lived Experience Experts. The new collaborative will place Lived Experience Experts at the center of research. What research is done, how it is done, and how the results are used and shared must be decided in partnership with lived experience expertise. The collaborative must also advance health equity for all. Every initiative and project must improve diversity, inclusion, and belonging.This paper proposes infrastructure and workforce development processes for the new research collaborative. Recommendations were developed by groups of clinical, research, lived experience, and health equity experts. The groups made sure everyone was able to contribute meaningfully and confidently. Every voice was heard and valued. This is also how the collaborative must operate, with shared leadership and teams that are trained to work well together. Education and processes are proposed to develop a diverse, inclusive interdisciplinary workforce, reflecting the community it serves and integrating lived experience expertise throughout. The proposed infrastructure is a network of expertise, resources, facilities, and processes, all connected by a platform. It is designed to start small, with just enough of each component to support a few simple projects. All projects will be evaluated to learn what works well and what can be improved. The whole collaborative will improve with learnings from each success and shortcoming.

Rare diseases, defined as conditions with a prevalence of less than one in 2000, are a public health priority, as they are collectively common and associated with poor outcomes. General practitioners (GPs) are central to improving outcomes for people living with rare disease by providing holistic, lifelong and family-centred care. However, the complexity, number of rare diseases and system-wide limitations create challenges for GPs and patients alike. This article outlines the challenges faced by people living with rare disease and highlights supportive resources for GPs to provide, facilitate and coordinate care. The new National Recommendations for Rare Disease Health Care (the Recommendations), co-designed with people living with rare disease, provide practical guidance on supporting patients on their journeys. Using a case study, we show how GPs can use the Recommendations.

Bird feathers serve numerous essential functions, with water repellency being particularly critical, as many other feather functions depend on remaining dry. Despite its importance, significant gaps remain in our understanding of the mechanisms underlying feather water repellency. This property arises primarily from feather microstructure. Feather morphology varies widely among bird species. However, the functional consequences of this variation for feather wettability remain poorly understood. Advances in computer-aided 3D modeling enable us to investigate such structure-function relationships with unprecedented precision. In this study, we created 3D feather models from micro-computed tomography scans of a Seaside Sparrow feather and systematically altered its morphology, specifically barbule density, to observe how feathers interact with water droplets using computational fluid dynamics simulations. We quantified the spatio-temporal distribution of droplet behavior, measured spreading radii during interactions, and assessed final contact angles for feathers with varying barbule densities. We validated our simulation results with barbule-density estimates and static contact-angle measurements on contour feathers from four New World sparrow species, including the Seaside Sparrow, spanning a gradient in water exposure. Together, computational and experimental results demonstrate that higher barbule density reduces water repellency. Collectively, these findings highlight how feather microstructure directly governs feather wettability.

Pediatric pulmonary arterial hypertension (PAH) carries high mortality with 81% 5-year transplant-free survival. The global Tracking Outcomes and Practice in Pediatric Pulmonary Hypertension-2 (TOPP-2; NCT02610660) registry was created to assess the treatments and outcomes of newly diagnosed pediatric PAH patients. This study evaluates real-world treatment strategies and their relationship with outcomes. Within TOPP-2, 445 subjects with newly diagnosed, catheterization-confirmed WSPH Group 1 pediatric PAH were enrolled. Treatment regimens were classified as none, calcium channel blocker monotherapy, PAH-targeted Monotherapy (Mono), Dual, Triple (enteral/inhaled only), or triple including parenteral prostanoid (TripleX). Baseline treatment strategy was defined as medications received three months following diagnosis. Primary clinical endpoint was death or lung transplantation. Dual therapy was the most common baseline treatment regimen (40.2%), followed by Monotherapy (29.0%). Phosphodiesterase type 5 inhibitors were the most common class of PAH-targeted therapy (72.4%) followed by endothelin receptor antagonists (59.3%). Adjusting for disease severity at diagnosis, baseline Dual patients had lesser hazard of death/transplant than Mono patients escalating to Dual by year one (HR=0.30, 95% CI=0.16-0.56, p<0.001). Baseline TripleX patients had lesser hazard of death/transplant than those started on enteral/inhaled therapy only and escalated to parenteral by year one (HR=0.28, 95% CI=0.15-0.50, p<0.001). A wide range of pediatric PAH initial medication strategies were observed in the TOPP-2 registry. Therapy regimen escalation within the first year, to Dual for lower-risk patients or to TripleX for higher-risk patients, was associated with worse outcomes compared to those treated more aggressively upfront, supporting upfront over sequential combination therapies.

There is a deficit in training, skill development opportunities, and resources needed to prepare Black early-career scholars for sustainable careers in gerontology. Historically Black Colleges and Universities (HBCUs) play an important role in this effort, as they produce a large percentage of Black graduates in social science and health fields. This paper describes one initiative designed to diversify the gerontological workforce: the HBCU Aging Conference. Co-organized by the Gerontological Society of America's HBCU Collaborative Interest Group and Blacks in Gerontology and Geriatrics, the first convening occurred as a soft launch in 2023, followed by the inaugural full conference in 2024. Based on a succinct review of conference activities, this paper presents recommendations for preparing future scholars to address knowledge gaps in Black aging research. The HBCU Aging Conference has created a critical pipeline for training and supporting the next generation of Black scholars in aging.

Bright Futures Parent and Patient Handouts from the American Academy of Pediatrics are widely used after pediatric well-child visits, yet their content has not been systematically evaluated. We assessed all fourth-edition handouts using standard readability formulas for readability and the Patient Education Materials Assessment Tool for Printable Materials (PEMAT-P) for understandability and actionability. Patient Handouts averaged a 4th-grade level (ages 7-8), 5th-grade level (ages 9-10), and 6th- to 7th-grade level (age 11+). Parent Handouts ranged from about 5th to nearly 7th grade. The PEMAT-P scores across all 23 handouts showed high understandability (92.3%) but lower actionability (60%). Overall, Bright Futures Handouts appear readable and understandable, supporting their use in pediatric practice, although opportunities remain to improve actionability. When clinicians create or evaluate patient education materials, attention to plain language, straightforward delivery, visual aids, and checklists helps improve readability, understandability, and actionability. These strategies may support more effective patient education and promote pediatric health outcomes.

While defocus incorporated multiple segments (DIMS) spectacle lenses are clinically proven effective in slowing myopia progression, the long-term, real-world patient experience is poorly understood. This study aimed to provide both quantitative and qualitative analyses of the wearer experience in a long-term cohort of DIMS wearers. This retrospective, convergent parallel mixed-methods study included 63 former child participants (now aged 17-23 years) who participated in the previous DIMS spectacle lenses clinical trial 8-10 years earlier. Data were collected 2 years after the final clinical trial concluded. A quantitative online questionnaire assessed wearer experience, while in-depth, semi-structured phone interviews explored the lived wearer journey. Predictors of satisfaction were identified using multiple linear regression and qualitative transcripts were analysed using a systematic thematic approach. Quantitative analysis revealed high satisfaction, with 88.9% of wearers satisfied or very satisfied and 90.4% believing in the efficacy of DIMS spectacle lenses. Regression analysis identified Maintenance and Durability as the strongest predictors of long-term satisfaction, alongside positive perceptions of appearance (adjusted R2&#x2009;=&#x2009;0.50, p&#x2009;<&#x2009;0.001). Qualitative analysis revealed a patient journey defined by parent-driven initiation, a manageable sensory adaptation period and strong perceived efficacy based on personal experience. However, only one-third of participants had continued DIMS use; a decision driven overwhelmingly by the high cost of the spectacles. DIMS spectacle lenses provide a highly positive long-term wearer experience. However, the high cost of the spectacles creates a dominant socioeconomic barrier to treatment continuity, highlighting a critical gap between clinical success and access outside the trial setting.

Uniportal non-coaxial spinal endoscopic surgery (UNSES) via far-lateral approach (FLA) is an innovative minimally invasive procedure for lumbar degenerative diseases, particularly far-lateral disc herniation and foraminal stenosis. However, complex lateral lumbar anatomy and strict endoscope-instrument coordination create a distinct learning curve that may compromise early surgical efficiency and safety. This study aimed to evaluate the efficacy and safety, quantify the learning curve, and to provide clinical guidance for the standardized promotion and application of this technology. A total of 40 consecutive patients with lumbar degenerative diseases who underwent UNSES via FLA by a single surgeon between January 2025 and December 2025 were included. All data were analyzed using SPSS 26.0 statistical software (IBM, USA). Primary outcomes included operation time, blood loss, fluoroscopy frequency, and intraoperative complication rate. Secondary outcomes were VAS, ODI, and modified Macnab criteria at 1, 3, and 6&#x2009;months postoperatively. The learning curve and the inflection point of the learning curve was determined using cumulative sum (CUSUM) analysis. The differences in clinical indicators between early and proficient stage were compared. Operation time, blood loss, and fluoroscopy times decreased significantly with case accumulation (p&#x2009;<&#x2009;0.05). CUSUM identified an inflection point at the 16th case, after which operation time stabilized at (55.3&#x2009;&#xb1;&#x2009;8.6) min, much shorter than the early phase (89.5&#x2009;&#xb1;&#x2009;10.3) min (p&#x2009;<&#x2009;0.001). Before the 16th case, the curve was in an upward trend; after the 16th case, the curve tended to be flat, indicating the proficiency stage. Postoperative VAS and ODI improved significantly than those before surgery at each follow-up time (p&#x2009;<&#x2009;0.05). There was no significant difference in postoperative VAS score and ODI between the two groups at each follow-up time point (p&#x2009;>&#x2009;0.05). The total complication rate was 12.5% (5/40), were cured by conservative treatment. The total excellent-good rate was 90.0% (36/40). L5/S1 and Bertolotti's syndrome were independent factors affecting the learning curve. UNSES via FLA is a safe and effective minimally invasive technique for treating complex lumbar degenerative diseases. It has a certain learning curve, and the inflection point is about the 16th case. After mastering the key techniques such as anatomical positioning, endoscopic manipulation and hemostasis, the surgeon can gradually reach the proficiency stage, with significantly improved surgical efficiency and clinical efficacy, and controllable complications. This study provides a theoretical basis for the clinical training and technology promotion of UNSES via FLA.

Bacteria are frequently attacked by viruses, known as phages, and rely on diverse defence systems to survive. While phages can evade defences by covalently modifying their DNA, these non-canonical nucleobases create molecular signatures that bacteria can exploit. Here, using structure-guided discovery, we identified two widespread families of anti-phage DNA glycosylases, Dag1 and Dag2. Although DNA glycosylases are classically associated with DNA repair, Dag1 and Dag2 act as antiviral effectors that selectively target phages carrying modified guanine bases. Guided by the conserved glycosylase fold, we uncovered numerous defence-associated glycosylases that collectively form a diverse repertoire of enzymes targeting chemically modified phage DNA. We further identified a distinct glycosylase superfamily that protects against phages carrying modified thymidine bases. Together, these findings establish DNA glycosylases as a versatile class of bacterial immune proteins and highlight structure-guided discovery as a powerful strategy for uncovering hidden antiviral defences.

Real-world evidence (RWE) plays an expanding role in regulatory, health technology assessment (HTA), and lifecycle decision-making, prompting a rapid increase in guidance documents intended to support its generation and use. This commentary argues that additional guidance is not redundant; rather, it is necessary to sustain consistency, credibility, and confidence as RWE methods become more specialized and operationally complex. Recent advances, including pragmatic and registry-based trials, hybrid randomized-real-world designs, external control arms, artificial intelligence and machine learning applications, digital health data, synthetic controls, and data tokenization, have outpaced the scope of many existing frameworks. Although recent reporting initiatives and international harmonization efforts have improved transparency and reproducibility, important gaps remain in implementation-focused guidance on issues such as causal inference, dynamic borrowing, linkage validation, algorithm auditability, reproducibility, and distributed data environments. The need is further amplified by heterogeneity across regulatory and HTA agencies, where differing evidentiary expectations can create uncertainty and inefficiency, and by the limited availability of context-appropriate guidance for low- and middle-income countries, where structural data and infrastructure constraints may hinder RWE generation and use. Future progress should emphasize targeted, modular, and potentially "living" guidance that is updated as methods evolve, while also improving uptake of existing frameworks through clearer reporting expectations, education, and stakeholder collaboration. More guidance, when focused and implementation-oriented, can better translate methodological principles into decision-grade RWE.

This survey explores how machine learning and artificial intelligence (AI) can be integrated with mechanistic models to create more accurate, dynamic, predictive, and personalized representations of biological systems, commonly referred to as digital twins (DTs). Mechanistic models, such as pathway-based Boolean or differential equation frameworks, provide interpretable insights into biological processes; however, calibrating these models to represent individual variability across large, heterogeneous cohorts remains a significant challenge, as their physically constrained structures often lack the flexibility to capture complex, non-mechanistic nuances in patient data. Focusing on elderly cancer patients-a vulnerable population underrepresented in clinical research-we discuss how hybrid DTs can bridge the gap between interpretable mechanistic frameworks and flexible, predictive AI approaches, enabling continuous monitoring, risk stratification, and adaptive treatment planning. To illustrate these principles, we present a proof-of-concept case study involving a synthetic breast cancer dataset in which comprehensive geriatric assessment, clinical tests, and quality of life measures inform dosing decisions for older patients via a Markov Decision Process. By combining a synthesis of current literature with the application of a sequential decision-making framework optimized using longitudinal data, this work provides a foundational understanding for researchers and clinicians interested in leveraging DTs to improve personalization and outcomes in geriatric oncology.

The accuracy of predicting ice phase changes on transmission lines is influenced by the interaction of various meteorological factors. However, existing models, which ignore the interactions between multiple physical fields, result in insufficient prediction accuracy and other issues. To address this, a multi-physical field coupling ice phase change prediction model has been developed. First, by setting up multi-physical field coupling boundary conditions, the model simulates various meteorological conditions to create a three-dimensional geometric model of ice phase changes on transmission lines. Then, different kernel functions are selected to verify the reliability of the support vector machine (SVM) prediction algorithm. Finally, the multi-physical field coupling model is used for ice phase change prediction experiments, and its effectiveness is verified through comparison with actual measurement data and SVM prediction data. The experimental results show that the model has an average prediction accuracy of 98.12% and an average precision of 98.54%, significantly improving upon traditional methods. This model provides high-precision decision support for the early warning and protection of ice disasters on transmission lines, making it highly valuable for engineering applications.