HAL Portal Univ-Côtedazur
Not a member yet
95391 research outputs found
Sort by
Depth- and Distance-Dependent Correlated Seismic Noise Analysis at LSBB and Implications of Newtonian Noise Mitigation for the Einstein Telescope
No full textInternational audienceCorrelated seismic noise and the resulting Newtonian noise (NN), pose a significant challenge for future gravitational-wave detectors like the Einstein Telescope (ET), especially for low-frequency gravitational-wave background (GWB) searches.The body-wave-induced NN could significantly impact ET's sensitivity to GWB by orders of magnitude in its most sensitive frequency band.This study characterizes correlated seismic noise over varying horizontal separations and depths at the Laboratoire Souterrain à Bas Bruit (LSBB) to assess this impact and inform about NN mitigation strategies as well as potential design choices (e.g. exact depth of the facility).The seismic data is being collected utilizing three mobile seismometers alongside three fixed LSBB sensors to capture seismic field fluctuations at distances exceeding 250 m.In this talk, we discuss results from the analysis that focuses on horizontal separations of approximately 275 m, 550 m, and 825 m, while also evaluating depth-dependent effects.Coherence and cross-spectral density (CSD) analyses reveal strong seismic correlation below ~2 Hz, with significant correlation persisting up to ~40 Hz, particularly at the shortest (~275 m) separation relevant for ET's inter-detector distances.This finding provides confirmation of the correlated underground seismic noise, which can be used to predict the resulting correlated NN.These LSBB site specific results quantify the challenge posed by correlated seismic noise for ET's low-frequency GWB searches and provide essential input for developing and validating the effective NN cancellation strategies vital for ET
Value of non-invasive test dynamics in guiding HCC surveillance decisions after HCV cure in patients with cirrhosis
No full textInternational audienceBackground & aims: Whether the dynamics of non-invasive tests (NITs) correlate with hepatocellular carcinoma (HCC) risk in patients with cirrhosis following sustained virological response (SVR) remains unknown. Thus, we aimed to describe NIT dynamics and assess their correlation with HCC risk.Methods: The dynamics of NITs (fibrosis-4 index [FIB-4], aspartate aminotransferase-to-platelet ratio index [APRI] and liver stiffness measurement) were described in patients with cirrhosis after SVR included in two prospective French multicenter cohorts (ANRS CO22 Hepather and CO12 CirVir) between 2006 and 2015. To assess their relationship with the risk of HCC, a joint modeling approach was employed using both standard and flexible models adjusted for age and sex. The impacts of NIT current value and slope during follow-up on HCC risk were assessed, considering competing risks of death.Results: A total of 3,067 patients with cirrhosis who achieved SVR were analyzed, among whom 228 (7.4%) developed HCC and 210 (6.9%) died during a 26-month follow-up. All NITs were increased at baseline in patients who ultimately developed HCC, whereas platelet counts were lower. All NITs improved in patients who did not develop HCC. More varied changes were observed during the follow-up of patients who ultimately developed HCC. Joint model analyses showed that current values of FIB-4, APRI and platelet count at any time impacted HCC risk. Only FIB-4 and APRI slopes influenced the same outcome. When considering NIT current value and slope simultaneously, only the current value of NITs impacted HCC risk while the slopes were not informative.Conclusions: The dynamics of NITs following SVR do not identify patients with cirrhosis who could be safely excluded from surveillance programs. NIT current value is more informative than slope, which will necessitate regularly re-assessment of HCC risk to design individualized surveillance strategies.Impact and implications: It has been postulated that monitoring non-invasive test (NIT) dynamics following HCV cure may provide information on the residual risk of hepatocellular carcinoma (HCC) in patients with cirrhosis and may allow for the discontinuation of surveillance in certain patient subsets. We analyzed data from over 3,000 patients and found that while all NITs improved in patients with cirrhosis who did not develop HCC, those who eventually developed liver cancer showed more varied changes in these tests. Specifically, the current values of tests like FIB-4 (fibrosis-4 index) and APRI (aspartate aminotransferase-to-platelet ratio index) were linked to an increased risk of HCC, while their slopes did not provide additional useful information, suggesting that dedicated prospective studies are warranted to define how repeated measurement of NITs could be combined with other variables into HCC risk stratification algorithms. Until then, HCC surveillance should be maintained in all patients with cirrhosis following HCV eradication, even in case of decreased NIT values
Causal Meta-Analysis: Rethinking the Foundations of Evidence-Based Medicine
No full textMeta-analysis, by synthesizing effect estimates from multiple studies conducted in diverse settings, stands at the top of the evidence hierarchy in clinical research. Yet, conventional approaches based on fixed-or random-effects models lack a causal framework, which may limit their interpretability and utility for public policy. Incorporating causal inference reframes meta-analysis as the estimation of well-defined causal effects on clearly specified populations, enabling a principled approach to handling study heterogeneity. We show that classical meta-analysis estimators have a clear causal interpretation when effects are measured as risk differences. However, this breaks down for nonlinear measures like the risk ratio and odds ratio. To address this, we introduce novel causal aggregation formulas that remain compatible with standard meta-analysis practices and do not require access to individual-level data. To evaluate real-world impact, we apply both classical and causal meta-analysis methods to 500 published meta-analyses. While the conclusions often align, notable discrepancies emerge-revealing cases where conventional methods may suggest a treatment is beneficial when, under a causal lens, it is in fact harmful
On leveraging browser-level measurements for network troubleshooting
No full textInternational audienceDespite advances in network technologies, slow web browsing remains a persistent issue, and its troubleshooting continues to be challenging. To address this challenge, our work presents a lightweight passive measurement solution that relies on web performance data readily available within the browser, such as the connect time, the response time, and the page load time, to infer network performance, detect anomalies, and troubleshoot their origins. Through controlled network experiments with manually injected anomalies, including multiple concurrent performance issues, and leveraging a dataset of more than 43, 000 webpages and tens of thousands of network scenarios, we develop a predictive model using machine learning that is capable of estimating network performance metrics with acceptable accuracy. By solely relying on users' daily web activity, our solution can continuously monitor network performance, identify anomalies, and provide actionable insights, without overloading the network with measurement probes, thus making network troubleshooting accessible and non-intrusive for everyday users
Reconstruction of inclined extensive air showers using radio signals: from arrival times and amplitudes to direction and energy
No full textInternational audienceRadio detection is now an established technique for the study of ultra-high-energy (UHE) cosmic rays with energies above eV. The next-generation of radio experiments aims to extend this technique to the observation of UHE earth-skimming neutrinos, which requires the detection of very inclined extensive air showers (EAS). In this article we present a new reconstruction method for the arrival direction and the energy of EAS. It combines a point-source-like description of the radio wavefront with a phenomenological model: the Angular Distribution Function (ADF). The ADF describes the angular distribution of the radio signal amplitude in the 50-200 MHz frequency range, with a particular focus on the Cherenkov angle, a crucial feature of the radio amplitude pattern. The method is applicable to showers with zenith angles larger than , and in principle up to neutrino-induced showers with up-going trajectories. It is tested here on a simulated data set of EAS induced by cosmic rays. A resolution better than 4 arc-minutes () is achieved on arrival direction, as well as an intrinsic resolution of 5% on the electromagnetic energy, and around 15% on the primary energy
High-Dimensional Variable Clustering based on Maxima of a Weakly Dependent Random Process
No full textInternational audienceWe propose a new class of models for variable clustering called Asymptotic Independent block (AI-block) models, which defines population-level clusters based on the independence of the maxima of a multivariate stationary mixing random process among clusters. This class of models is identifiable, meaning that there exists a maximal element with a partial order between partitions, allowing for statistical inference. We also present an algorithm depending on a tuning parameter that recovers the clusters of variables without specifying the number of clusters a priori. Our work provides some theoretical insights into the consistency of our algorithm, demonstrating that under certain conditions it can effectively identify clusters in the data with a computational complexity that is polynomial in the dimension. A data-driven selection method for the tuning parameter is also proposed. To further illustrate the significance of our work, we applied our method to neuroscience and environmental real-datasets. These applications highlight the potential and versatility of the proposed approach. Supplementary materials for this article are available online, including a standardized description of the materials available for reproducing the work
From the material to the immaterial: digital comparative anatomy for archaeological ungulates
No full textInternational audienceContext The identification of faunal remains (bones and teeth) found in archaeological contexts is based on a visual comparison of the shape and size of the bones with their modern counterparts. This comparative anatomy work draws on several morphological criteria identified for given species in present-day environments. However, the application of anatomical criteria by archaeozoologists presents two significant challenges. Firstly, while most of the criteria can be used to identify species, some are present in other species within the same genus or family, which limits identification and results in a significant loss of information. Secondly, there are discrepancies between the anatomical description of the criterion and its observation on the archaeological find. These discrepancies may be due to imprecise vocabulary, a lack of anatomical knowledge or subjectivity in the anatomical representation. This can lead to errors in anatomical identification.Main argumentIn light of these considerations, we sought to ascertain how these anatomical shapes could be described and observed using 3D imaging and supervised learning algorithms. The objective is to create a reliable and reproducible digital reference database based on expert knowledge. While the combination of topological data analysis and kernel matrices has yielded promising results in differentiating between wild and domestic species with similar morphological characteristics (Vuillien et al., 2024), the lack of a defined benchmark hinders the interpretation of the results from a morphological perspective. Furthermore, the construction of a digital database necessitates the utilisation of numerically discernible anatomical criteria. This prompted us to ask the following questions: We would like to ascertain whether the anatomical criteria observed on the bones are identical to those observed on the 3D scanned part (Figure 1). Are there specific criteria that apply only to the 3D object? If so, can these newly identified criteria be considered reliable identification keys?ApplicationsTo address these questions, we collaborated with archaeozoologists and mathematicians on a data set comprising 150 bones from five ungulate species with similar morphological characteristics: ibex, goat, sheep, roe deer and gazelle. While some of these species, such as sheep and goats, have a substantial literature supporting their identification, others, such as roe deer and gazelle, lack sufficient evidence. This work has enabled us to update the available anatomical criteria and to identify and test numerical criteria that are not visible on physical bones. This work forms the basis of future machine learning analyses that will enable us to consolidate the identifications of animals found in archaeological contexts with the help of the expert and to propose new identification criteria that are more mathematically robust
The Deep Zero-Inflated Latent Position Block Model for the Clustering of Nodes in Graphs
No full textNational audienceThe evolution in storage capacities has led to a data explosion, making networks essential for modeling relationships between objects (nodes). These complex networks require effective clustering and visualization methods to summarize and inter- pret their information. The deep latent position block model (Deep-LPBM), designed for binary networks, combines partial block-based clustering and continuous latent representation to visualize nodes. Here, we propose an extension, the deep zero-inflated latent position block model (Deep-ZLPBM), designed for non-binary networks, where the entries of the adjacency matrix can take integer values. This model is based on a deep variational autoencoder that integrates a graph convolutional network (GCN) and a decoder leveraging a zero-inflated Poisson (ZIP) distribution. Inference relies on the maximization of the marginal likelihood, and optimization is performed using stochastic gradient descent.L'augmentation des capacités de stockage a entraîné une explosion des données, rendant les réseaux essentiels pour modéliser les relations entre objets (noeuds). Ces réseaux, souvent complexes, nécessitent des méthodes efficaces de clustering et de visualisation pour en résumer l'information. Le deep latent position block model (Deep-LPBM), conçu pour les réseaux binaires, combine clustering par blocs et représentation latente continue pour visualiser les noeuds. Nous proposons ici une extension, le deep zeroinflated latent posistion block model (Deep-ZLPBM), adapté aux réseaux non binaires, où la matrice d'adjacence peut prende des valeurs entières. Ce modèle repose sur un auto-encodeur variationnel intégrant un réseau de convolution de graphes (GCN) et un décodeur utilisant une loi ZIP (zero-inflated Poisson). L'inférence se concentre sur la maximisation de la vraisemblance marginale, et l'optimisation s'effectue par descente de gradient
Share the Ride: The Determinants of Long-Distance Carpooling Pricing Strategies in France
No full textInternational audienceThis paper investigates the pricing strategies used in long-distance carpooling in France. We investigate how several factors affect carpooling prices using a comprehensive dataset of BlaBlaCar trips combined with sociodemographic and intermodal competition data. The analysis identifies two distinct pricing patterns within the platform: one characterized by standardized and consistent pricing, and another marked by more flexible, market-responsive price setting. By focusing on price per minute, we examine how trip characteristics, competitive conditions, and demand heterogeneity affect these pricing behaviors. The results show that variables such as the number of stopovers, trip length, airport or cross-border connections, and the presence of alternative transport modes influence pricing, but with contrasting effects across the two patterns. The standardized approach tends to reflect cost-sharing principles and reinforces network effects, while the more flexible approach adapts dynamically to local competition and demand
Thyroidectomy without radioiodine in patients with low-risk thyroid cancer: 5 years of follow-up of the prospective randomised ESTIMABL2 trial
No full textInternational audienceBackground ESTIMABL2, a multicentre randomised phase 3 trial in patients with low-risk differentiated thyroid cancer (ie, pT1am or pT1b, N0 [no evidence of regional nodal involvement] or Nx [involvement of regional lymph nodes that cannot be assessed in the absence of neck dissection]), showed the non-inferiority of a follow-up strategy without radioactive iodine (¹³¹I) administration compared with a postoperative ¹³¹I administration at 3 years post-randomisation. Here, we report a pre-specified analysis after 5 years of follow-up.Methods Patients treated with total thyroidectomy with or without prophylactic neck lymph node dissection, without postoperative suspicious findings on neck ultrasonography, were randomly assigned to the no-radioiodine group or to the radioiodine group (1•1 GBq-30 mCi after recombinant human thyrotropin-stimulating hormone). Follow-up consisted of annual thyroglobulin and thyroglobulin antibody determinations during levothyroxine treatment and neck ultrasonography in odd-numbered years. An event was defined as abnormal foci of ¹³¹I uptake on the post-treatment whole-body-scan requiring subsequent treatment, abnormal neck ultrasonography, elevated thyroglobulin levels, increasing titres or appearance of thyroglobulin antibody (using the same laboratory assay), or a combination of these definitions. Non-inferiority of the proportion of patients without an event in one group compared with the other at 5 years after randomisation was shown if this proportion and its CI did not differ by more than -5%. This study was registered on ClinicalTrials.gov (NCT01837745) and is completed. FindingsOf the 776 patients (n=642 [82•7%] female and n=134 [17•3%] male, median age 52•9 years [IQR 42•6-63•1]) enrolled, 698 were evaluable at 5 years. The proportions of patients without events were 93•2% in the no-radioiodine group and 94•8% in the radioiodine group, for a difference of -1•6% (90% CI -4•5 to 1•4). Events consisted of structural or functional abnormalities (n=11) and biological abnormalities (n=31).Interpretation The non-inferiority of a follow-up strategy compared with postoperative ¹³¹I administration in low risk differentiated thyroid cancer was confirmed at 5 years. There is no loss of opportunity in following these patients without postoperative ablation.Funding Programme de Recherche Hospitalier Clinique.</div