136986 research outputs found
Sort by
Author Correction: Functional mechanisms underlying pleiotropic risk alleles at the 19p13.1 breast–ovarian cancer susceptibility locus (Nature Communications, (2016), 7, 1, (12675), 10.1038/ncomms12675)
No full textThe psychometric performance of the EQ-HWB-9 for measuring health and wellbeing in a general population sample from Australia and New Zealand
No full textPURPOSE: The EuroQol Health and Wellbeing Short (EQ-HWB-9) is a new, generic 9-item instrument, suitable for evaluating interventions in health and social-care settings for patients and caregivers. The instrument now requires validation across general and caregiver populations. Informal caregiving can be time-intensive and impact caregiver's physical and mental well-being. However, caregiver outcomes are often overlooked in healthcare decisions, which can lead to inefficient resource allocation. We aimed to examine the psychometric performance of the EQ-HWB-9 in a general population dataset, including caregivers of persons with disability/chronic illness. METHODS: Using general population samples, stratified by age, gender, region, ancestry, and income for Australia and New Zealand, we investigated EQ-HWB-9 item distribution and known-group validity (t-tests; Cohen's d for effect size, with sub-group analysis by country, gender and age) across sum-scores and UK pilot preference-weighted scores. Item scores were compared across caregiver groups. Convergent validity was assessed between the EQ-HWB-9 and the Kessler-6 using Spearman's Rho. RESULTS: The sample included 2542 participants, 2018 from Australia and 524 from New Zealand. Item distribution was similar to previous studies. Known-group validity results aligned to a priori hypotheses for caregiver, mental health, physical health and disability and sleep issues variables. Caregivers had significantly higher scores across each item than their counterparts. Convergent validity conformed to a priori expectations. CONCLUSION: The EQ-HWB-9 appears valid in this general population setting. This study helps to build the evidence for the use of the instrument across diverse settings. Australian- and New Zealand-specific value-sets would be a good future addition
Towards Schema-Aware Embeddings for Inductive Knowledge Graph Cardinality Estimation
No full text© 2025 Ran LiangCardinality estimation is the task to predict query result sizes without actually executing the queries. Estimation results can be used to guide query engines in selecting efficient execution plans. Query performance may get orders-of-magnitude degradation with large estimation errors. Recent neural approaches for knowledge graph cardinality estimation achieve high accuracy using learned embeddings and graph neural networks. They focus on transductive assumptions where all entities in test queries were observed during training. This assumption may not always hold in real-world scenarios that demand inductive generalization, where test queries for online model inference may contain entities unseen at training.
This thesis identifies and addresses two inductive scenarios. First, dynamic knowledge graphs continuously introduce new entities through user-generated content, scientific publications, or real-time feeds, where completely unseen entities lack pre-trained embeddings. Second, even in static knowledge graphs, computational constraints may prevent exhaustive entity coverage during cardinality estimation model training. Massive knowledge graphs such as Wikidata contain millions of entities, making comprehensive training query coverage infeasible. While entity embeddings may exist from separate pre-training, estimation models may not be able to learn their selectivity patterns in cardinality-specific contexts.
We propose SGCE, Schema-Guided Cardinality Estimation, an embedding model that leverages schema information (i.e., classes and relations) to address the two inductive scenarios. Our main observation is that while individual entities may lack training exposure, their semantic classes can provide useful regularization signals. For the first scenario, we use random embeddings augmented with class-level representations derived from averaging embeddings of training entities within each class. For the second scenario, we develop an attentive fusion mechanism that adaptively combines entity and class embeddings, learning to balance entity-specific features with class-level selectivity patterns. Furthermore, we designed relation-aware literal embeddings to enhance both approaches. Instead of treating all literals in a generic ``literal'' class, we differentiate them based on their associated relations, which are another core component of the knowledge graph schema.
We conduct experiments on star-shape and path-shape queries extracted from four different knowledge graphs. We evaluate accuracy in terms of q-error and show that SGCE achieves consistent cardinality estimation error reductions over state-of-the-art baselines. Meanwhile, SGCE maintains practical computational efficiency with small inference overhead. Through ablation studies, we observed that simple concatenation of entity and its class embeddings often shows larger error. The attentive fusion mechanism successfully addresses this problem with better performance. The further relation-aware literal embeddings also prove effective in most experiments.
Overall, this work contributes to query optimization for knowledge graph systems that need to handle previously unseen entities. By showing that class-aware embeddings provide crucial semantic regularization for generalizing beyond training entities, we expand the applicability of learned cardinality estimation models to dynamic and large-scale knowledge graphs where complete entity coverage during training is not practical
Neuroinflammation associated with proviral DNA persists in the brain of virally suppressed people with HIV
No full textDespite viral suppression with antiretroviral therapy (ART), people with HIV (PWH) continue to exhibit brain pathology, and ~20% of individuals develop HIV-associated neurocognitive disorders. However, the state of cellular activation in the brain of virally suppressed (VS) PWH and the impact of local viral reservoirs on cellular activation are unclear. Using multiplex immunofluorescence imaging, here, we demonstrate that the frontal cortex brain tissue from both non-virally suppressed (nVS; n=17) and VS PWH (n=18) have higher frequencies of astrocytes and myeloid cells expressing interferon-inducible Mx-1 and proinflammatory TNFα relative to HIV-seronegative individuals (p<0.05 for all). The frequency of TGF-β1+ cells were also elevated in the brain tissue from both nVS and VS PWH, which may support active immunoregulatory responses despite ART. Importantly, the frequency of Mx1+ myeloid cells correlated with levels of total HIV DNA and intact and 5' defective HIV proviral DNA (p<0.05 for all) in the brain of VS PWH. These findings demonstrate that cell activation persists in the brain of VS PWH and is associated with HIV DNA in the brain, which may contribute to neuropathology
Impact of stereotactic radiosurgery timing relative to immune checkpoint blockade administration on brain metastasis disease and radionecrosis outcomes
No full textBackground Whether stereotactic radiosurgery (SRS) and immune checkpoint blockade (ICB) for brain metastases (BrM) have a time window for synergistic efficacy and toxicity is unclear. We examined this question in a large, contemporary cohort of patients who received concurrent SRS and ICB. Methods Patients who received SRS for intact BrM within 1 month before to 6 months after an ICB cycle at a single center from 2018 to 2023 were included if they had no prior whole-brain radiotherapy or SRS to the same BrM. Intracranial progression-free survival (icPFS), local control (LC), distant brain control (DBC), overall survival (OS), and radionecrosis were analyzed by Kaplan-Meier and log-rank methods. Cox regression was used for uni/multivariable analysis (UVA/MVA). Results A total of 419 BrM, 170 treatment episodes, and 134 patients were analyzed. In total, 43% and 40% of patients had melanoma and non-small cell lung cancer. A shorter SRS-ICB interval significantly correlated with improved icPFS, LC, and OS, but not DBC. This was true when analyzed as either a categorical or continuous factor. On MVA, SRS-ICB interval outperformed all factors including histology, ICB type, and de novo BrM status in predicting icPFS (P=.030), LC (P=.042), and OS (P=.033). In the absence of corticosteroids, pre-SRS lymphocyte counts correlated with improved LC (P=.02). Radionecrosis was not associated with SRS-ICB interval, but with BrM size and number of ICB cycles received prior to SRS. Conclusion Delivering SRS closer to ICB cycles was associated with improved icPFS, LC, and OS without affecting radionecrosis rates. This may present a therapeutic opportunity to improve BrM outcomes
Post-traumatic stress disorder in older adults: a global collaboration on setting the future research agenda
No full textPost-traumatic stress disorder (PTSD) in later life poses a substantial burden on public health and social care systems. However, research in this population remains scarce. In this Personal View, we review the current state of research on PTSD and ageing, as presented by the On Traumatic Stress and Ageing: A Global Network task force, part of the Global Collaboration on Traumatic Stress. Evidence-based knowledge on PTSD in older (aged 60 years or older) trauma survivors was synthesised across four clinical domains: ageing mechanisms, assessment, treatment, and care. 142 publications were reviewed to integrate available evidence and establish consensus-based research priorities. The findings highlight the urgent need for high-quality research across all four domains on older trauma survivors. Future studies should focus on older under-represented groups, such as women; individuals with multiple comorbidities, including neurocognitive disorders; and populations in low-income and middle-income countries. Using standard diagnostic instruments, establishing clinically meaningful functional outcomes, and engagement of people with lived experience should be prioritised to be applied in future research
A synthesis of RCTs on psychological interventions fostering strengths and virtues: Evidence from 21 systematic reviews
No full textResearch on mental health advocates the cultivation of character strengths to enhance well-being. Existing meta-analyses support positive correlations between strengths and well-being, and an increasing number of randomized controlled trials (RCTs) have assessed the effects of interventions aimed at cultivating them. This pre-registered multiple systematic review of RCTs aims to map and synthesize evidence from these trials regarding the overall impact of such interventions on strengths and well-being. Based on Peterson and Seligman's Virtues in Action model, which groups 24 character strengths under 6 virtues as latent variables, we conducted searches for each character strength (21 in total), across four specialized databases, including gray literature. The included RCTs assessed the impact of virtue-based interventions with adult participants in studies including baseline and post-intervention measures of the targeted strength and well-being. After reviewing 61,479 abstracts, 162 RCTs were included representing 15 strengths grouped into the six virtues among 33,032 participants. In contrast, the interventions with the best evidence for increasing strengths and well-being were those that enhanced kindness, humor, hope, perspective, and gratitude. Our review revealed that there is a relative paucity of RCT studies in the field and a need to include robust outcome measures. While there is relatively robust evidence of the efficacy of RCTs to enhance certain character strengths and associated well-being, there is still insufficient evidence on the efficacy of most of the 24 strengths of Peterson and Seligman's model. We discuss implications of the findings and recommend guidelines for future research
3D Cadastral database implementation based on LADM edition II and IFC
No full textA major challenge in developing 3D cadastral databases is integrating legal spaces and physical elements within a 3D environment while aligning with international standards and connecting to external databases. This paper presents a database implementation based on LADM Edition II, integrated with the IFC standard to support both legal and physical aspects. The model links to authoritative spatial data sources in the Victorian cadastral system. The process covers conceptual, logical, and physical data models, addressing challenges during transformation and proposing solutions. It also outlines complexities in normalisation, data preparation, and the ETL process, including creating 3D legal spaces, extracting boundary faces, representing vertical extents such as airspace rights, and applying virtual boundaries. The database is evaluated through 3D spatial queries and visualisations using a real-world case study from Victoria, Australia. Findings demonstrate that the Victorian cadastral system complies with LADM Parts 1, 2, and 5, and provides guidance for implementing Part 6
Enhancing public health infectious disease response through the Australian Pathogen Genomics program (AusPathoGen): evaluation protocol
No full textINTRODUCTION: Pathogen genomics is rapidly becoming a cornerstone in the surveillance and response to infectious diseases. However, there is little evidence on how it shapes strategies for effective public health response and decision-making. This paper presents the evaluation protocol for the Australian Pathogen Genomics (AusPathoGen) program, which aims to assess the utility of whole genome sequencing in informing public health responses to infectious diseases in Australia. METHODS: A mixed methods approach will be adopted to systematically explore the utility of whole genome sequencing in public health action and decision-making through a series of linked projects. Methods include situation assessment surveys of Australian public health laboratories, expert elicitation, and case study analysis. The situation assessment surveys will gather data on public health laboratories' processes, practices, and associated costs for whole genome sequencing. Expert elicitation will seek views on the prioritization of pathogens for whole genome sequencing. Case studies of specific pathogens and outbreaks will serve as the basis for both impact assessment and qualitative comparative analysis. Genomic and epidemiological data will shed light on the influence of whole genome sequencing on outbreak response. DISCUSSION: This comprehensive evaluation of pathogen whole genome sequencing in Australia will enhance our understanding of how this data can be applied in public health response and decision-making. The methods discussed can be adapted to different public health pathogen genomic surveillance systems globally. Undertaking evaluation of such systems is crucial for identifying areas of improvement and providing recommendations to optimize quality, efficiency and resource allocation of pathogen genomics to improve public health responses
Dual-mobIlity verSus conventional Total hip arthroplasty In femoral Neck fractures, a registry-nested, open-label, Cluster-randomized crossover Trial (DISTINCT)
No full textAims A statistical analysis plan for analysis of DISTINC T (Dual-mobI lity verSus conventional Total hip ar throplast y In femoral Neck fractures, a registry-nested, open-label, Cluster-randomized crossover Trial) is outlined to aid transparency in analytical procedures. Methods DISTINCT aims to determine whether dual-mobility total hip arthroplasty (THA) is superior to conventional THA in reducing postoperative prosthetic hip dislocations in patients under going surgery for femoral neck fractures. The trial is nested within the Australian Orthopae dic Association National Joint Replacement Registr y (AOANJRR). The clusters comprised hospitals that per form at least 12 THAs for femoral neck fracture per annum. Each hospital was randomized to two sequential periods of dual-mobility THA and conventional THA, with the order randomly allocated. Crossover occurred once sites had recruited an average of 16 patients eligible for the primary analysis in the initial allocation, to provide an average of 32 patients per cluster. The primary outcome is the incidence of prosthetic hip dislocation in the first postoperative year. S econdary outcomes are the rate of revision for dislocation, all- cause revision, mor tality at one, two, and five years, and complication rates at one year. All analyses will use cluster summary methods, weighted by cluster size to account for unequal cluster sizes. Participants will be analyzed in the groups to which they were randomized (intention-to-treat). Sensitivity analyses will be per formed to analyze the unadjusted primar y outcome based on treatment received (as-treated) and per-protocol groupings. - - Conclusion This statistical analysis plan describes the detailed analysis of the DISTINCT study, a registr ynested, open-label, cluster-randomized crossover trial that will provide suppor ting evidence for prosthesis choice in the common clinical problem of femoral neck fracture