53417 research outputs found
Sort by
Domain-invariant representation learning of bird sounds
Get PDFPassive acoustic monitoring (PAM) is crucial for bioacoustic research, enabling non-invasive species tracking and biodiversity monitoring. Citizen science platforms provide large annotated datasets from focal recordings, where the target species is intentionally recorded. However, PAM requires monitoring in soundscapes, creating a domain shift between focal and passive recordings, challenging deep learning models trained on focal recordings. To address domain generalization, we leverage supervised contrastive learning by enforcing domain invariance across same-class examples from different domains. Additionally, we propose ProtoCLR, an alternative to SupCon loss which reduces the computational complexity by comparing examples to class prototypes instead of pairwise comparisons. We conduct few-shot classification based on BIRB, a large-scale benchmark to assess pre-trained bioacoustic models. Our findings suggest that ProtoCLR is a better alternative to SupCon
P-364. Efficacy and Safety by Sex Assigned at Birth After Switch to Doravirine/Islatravir (100 mg/0.25 mg) Once Daily: Week 48 Results from Two Phase 3 Randomized, Active-Controlled Studies in Adults Living with HIV-1
No full textCreating feedback with <scp>ImPACT</scp> : Improving consistency in feedback through the design of principles of good feedback and creating common <scp>QuickMarks</scp>
No full textAbstract Consistency in feedback is recognised as a crucial element for student assessment literacy and electronic feedback can facilitate this. The present project aimed to develop and refine common ways to communicate key feedback messages as part of transparency in assessment to enhance students' learning using QuickMarks (QMs; part of Turnitin) and support teachers' feedback literacy. Development of the QM sets required construction of good practice principles outlining what constituted good feedback: Improvement focus, Parsimony/Generality, Accessibility, Clarity and Consistency and Tone (acronym ImPACT). The project took place in two phases across two institutions. In both phases, psychology teaching staff were asked to provide either all or their perceived most important QMs. Content analyses identified topics fed back on, and these were used to develop QM sets using ImPACT. In phase 1, teaching staff reported use of common QMs facilitated marking efficiency and encouraged balance between positive and developmental feedback. Phase 2 used the same method to develop common QMs, but also aligned QMs to course learning outcomes in a different university. Student feedback was primarily positive and demonstrated acceptability and use of QMs, although students needed support to act on feedback. Our findings suggest common QMs can facilitate a continuing dialogue between teaching staff and students, improving transparency and supporting feedback literacy. Encouraging adherence to the use of the common set of QMs amongst teaching staff is still needed. Adopting ImPACT principles and common QMs could address challenges in education such as lack of transparency in feedback and enabling elements of personalisation in feedback. Context and implications Rationale for this study: The rationale for this study was to improve consistency in feedback thereby improving the transparency of the assessment process. Why the new findings matter: The paper describes a set of good practice principles that clarify what good feedback consists of and that developing and using a common set of feedback according to these principles is useful and acceptable to teaching staff and students. Implications for educational researchers and policy makers: This has implications to support teaching staff to develop their feedback literacy, and, in turn, support students to develop theirs. Led to the development of ImPACT principles, a set of guidelines that can be further tested to establish whether they describe effective feedback in Higher Education. </jats:p
Interactive Pathways of Key Prognostic Factors in Severe Asthma: A Bayesian Network Comparison of Clinical Trials & Real-World Data.
No full textBACKGROUND: The way in which risk predictors combine and contribute to severe asthma exacerbations may differ between clinical trials and real-world settings. RESEARCH QUESTION: How do the interactive pathways of risk predictors leading to severe asthma exacerbations compare under clinical trials versus real-world settings? STUDY DESIGN: and Methods: The analysis involved 345 severe asthma patients from the placebo arms of two international randomized control trails (RCTs), compared to 6814 biologic-naïve patients from the International Severe Asthma Registry (ISAR). Seventeen key risk predictors including demographics, biomarkers, lung function, healthcare use, exacerbation history, long-term oral corticosteroid use, asthma control, and nasal polyps were covered. The outcome was the occurrence of severe asthma exacerbations over 365-day following study enrolment. Bayesian Networks (BNs), obtained from machine learning combined with expert knowledge, elucidated significant interplay processes of risk predictors that led to severe asthma exacerbations. External validation was performed in each other cohort respectively. RESULTS: RCTs revealed 44 significant arcs (i.e., probabilistic inter-dependency) between 17 risk factors, while ISAR showed 170. Despite this difference, the main downstream prediction pathways were consistent across both settings, with two key pathways: total serum immunoglobulin E level influenced blood eosinophils to predict future severe exacerbations, and severe exacerbation history directly predicted future severe exacerbations. In external validation, RCTs -BN generalized well to ISAR patients (AUC = 0.68) whereas ISAR-BN underperformed in RCT patients (AUC = 0.50), while ISAR-BN demonstrated better calibration. INTERPRETATION: The core pathways predicting severe asthma exacerbations were similar in both RCTs and real-world settings, with comparable predictive performance
Deep learning to predict left ventricular hypertrophy from the electrocardiogram.
Get PDFAIMS: Left ventricular hypertrophy (LVH) is a strong predictor of cardiovascular disease. We previously compared supervised machine learning techniques to classify cardiac magnetic resonance (CMR)-derived LVH using ECG and clinical variables in 37,534 UK Biobank participants, obtaining an area under the receiving operating curve (AUROC) of 0.85, but with limited specificity and requiring external validation. In this study, we develop a deep learning (DL) model to improve classification with external evaluation in the Study of Health in Pomerania (SHIP). METHODS AND RESULTS: We analyzed 12-lead ECGs of 48,835 participants from the UK Biobank imaging study. The dataset was split into a training set (70%), validation set (15%) and test set (15%) for performance evaluation. The model architecture was a fully convolutional network, for which the input was the participants' median ECG and clinical variables and the predicted indexed left ventricular mass (iLVM) as the output. A subsequent logistic regression model was used to recalibrate iLVM predictions. In UK Biobank, 717 (1.5%) participants had CMR-derived LVH and the AUROC for the DL model was 0.97. The ECG components most predictive of LVH were the QRS complex and ventricular rate. The DL model outperformed our supervised algorithms, previous DL modelling efforts and clinical ECG benchmarks. There was modest generalizability of the DL model to 1,423 participants in SHIP (AUROC 0.78), with differences in clinical profile, ECG acquisition and CMR labelling as important factors. CONCLUSION: Our findings support the feasibility of scalable DL-based screening tools for prediction of LVH from the ECG, whilst highlighting the need for model development using larger datasets with greater diversity to ensure generalizability
A Large-Scale Genome-wide Association Study of Blood Pressure Accounting for Gene-Depressive Symptomatology Interactions in 564,680 Individuals from Diverse Populations.
No full textGene-environment interactions may enhance our understanding of blood pressure (BP) biology. We conducted a meta-analysis of multi-population genome-wide association studies of BP traits accounting for gene-depressive symptomatology (DEPR) interactions. Our study included 564,680 adults from 67 cohorts and 4 population backgrounds (African (5%), Asian (7%), European (85%), and Hispanic (3%)). We discovered seven previously unreported BP loci showing gene-DEPR interaction. These loci mapped to genes implicated in neurogenesis (TGFA, CASP3), lipid metabolism (ACSL1), neuronal apoptosis (CASP3), and synaptic activity (CNTN6, DBI). We also showed evidence for gene-DEPR interaction at nine known BP loci, further suggesting links between mood disturbance and BP regulation. Of the 16 identified loci, 11 loci were derived from non-European populations. Post-GWAS analyses prioritized 36 genes, including genes involved in synaptic functions (DOCK4, MAGI2) and neuronal signaling (CCK, UGDH, SLC01A2). Integrative druggability analyses identified 11 druggable candidate gene targets linked to pathways involved in mood disorders as well as known antihypertensive drugs. Our findings emphasize the importance of considering gene-DEPR interactions on BP, particularly in non-European populations. Our prioritized genes and druggable targets highlight biological pathways connecting mood disorders and hypertension and suggest opportunities for BP drug repurposing and risk factor prevention, especially in individuals with DEPR
Deep learning-based plaque characterization in hybrid IVUS-OCT images is superior to single-modality deep learning analysis and human experts: head-to-head comparison against histology.
No full textAIMS: Hybrid intravascular ultrasound-optical coherence tomography (IVUS-OCT) can enable more accurate plaque characterization than single-modality intravascular imaging, enhancing treatment planning and vulnerable plaque detection. However, image interpretation in IVUS-OCT is challenging and time-consuming. To overcome this limitation, we introduce a novel histology-trained deep learning (DL)-classifier for plaque component classification in IVUS-OCT images and compare its performance against single-modality DL and expert analysts. METHODS AND RESULTS: IVUS-OCT frames and matched histological sections from 10 cadaveric human hearts were included in this analysis. The histological data were used to define fibrotic, calcific, and necrotic core tissue regions of interest (ROIs) in IVUS-OCT and used to train three DL-classifiers for IVUS, OCT, or hybrid IVUS-OCT image analysis (992 frames) and test their performance (264 frames). The test set was additionally annotated by experts from three different core labs, and their estimations and those of the DL-classifiers were compared with histology.The IVUS-OCT DL-classifier had a superior performance to the IVUS-DL, OCT-DL, and the expert analysts in detecting plaque phenotypes (Kappa 0.60 vs. 0.19, 0.35, and 0.53, respectively) and accurately classified 68% of histologically defined fibroatheromas. The hybrid IVUS-OCT DL-classifier also had a better performance than single-modality DL-classifiers and the experts in assessing tissue types in ROIs annotated by histology (overall accuracy 86.7% compared with 73.2% for IVUS-DL, 66.6% for OCT-DL, and 70.6% for the experts). CONCLUSION: Plaque characterization using a histology-trained hybrid IVUS-OCT DL-classifier is feasible and enables more accurate detection of plaque components and phenotype classification than single-modality DL-classifiers and expert analysts
Pre-Diagnostic Features of Multiple Sclerosis in a Diverse UK Cohort: A Nested Case-Control Study.
Get PDFBACKGROUND: Many patients with Multiple Sclerosis (MS) experience nonspecific symptoms prior to diagnosis. This period-the 'MS prodrome'-has been described in socio-economically homogeneous cohorts to date. It remains unclear to what extent events prior to an MS diagnosis differ according to social determinants of health. METHODS: We conducted a retrospective, longitudinal, population-based nested case-control study using data from Clinical Practice Research Datalink (CPRD) Aurum. Associations between pre-diagnostic symptoms and MS risk were evaluated using multivariable logistic regression models in MS cases and matched controls. To determine whether associations differed by potential health determinants, we tested for statistical interactions and used stratified models. RESULTS: The study population consisted of 15,029 patients with MS (median index age 44.6, 80.3% female) and 81,027 age-matched controls. In the 5 years preceding diagnosis, MS cases were more likely than controls to have coded autonomic (OR 1.87 [1.80-1.94]), cognitive (OR 2.57 [2.06-3.20]), neurological (OR 7.91 [7.58-8.26]), pain (OR 2.21 [2.12-2.29]) and psychiatric (OR 1.75 [1.69-1.82]) symptoms. The direction of effect was consistent across gender, ethnicity, deprivation and location (urban vs. rural) strata, with over-representation of neurological symptoms in males and those living in urban areas. No statistically significant interaction between factors was found. CONCLUSION: We report a consistent relationship between the occurrence of prodromal symptoms and MS risk across a diverse UK cohort. No associations were specific to a particular ethnic, gender or socio-economic group. These findings strengthen the concept of pre-diagnostic symptoms reflecting a potential opportunity to identify those at the earliest stages of MS
Revolution of the Heartlands: A Comparative Study of Radical Conservative Thought in the United States and Hungary
No full textIn recent years, a friendship has blossomed between Donald Trump’s GOP and Viktor Orbán’s Fidesz. Their alliance is neither a mere publicity stunt, nor the product of the Hungarian regime’s malign influence. It is an illustration of how the conservative movements of both countries have radicalized in similar ways. In the 1990s and 2000s, radical conservative thinkers worked below the surface of respectable politics to develop complex critiques of liberal hegemony. In the United States, paleoconservatives like Samuel Francis and Paul Gottfried theorized that a new class of liberal managers and ideologues had captured American society through an elaborate administrative state. They called for a conservative revolution, which would dismantle that state and reassert the power of Middle America, the nation’s white middle-class heartlands. Similarly, in Hungary, national radicals like István Csurka and Gyula Tellér asserted that a new class of liberal managers and ideologues had subordinated their nation to an administrative apparatus directed by Western powers. They too advocated for a conservative revolution that would do away with that apparatus and reinstate the power of Central Europe, the continent’s own white middle-class heartlands. This dissertation reconstructs and compares the worldviews of the paleoconservatives and national radicals. It examines their arguments through four thematic lenses – fusionism, new class theory, metapolitics and anti-imperialism. Additionally, it demonstrates how Trump’s GOP and Orbán’s Fidesz have adopted the ideas of the radical right. Their assault on the bureaucratic organs of Washington, D.C., and Brussels, and on neoliberal ideology, are manifestations of a decades-long, intellectual project of conservative revolution