HAL Portal Univ-Côtedazur
Not a member yet
95391 research outputs found
Sort by
Motifs des évacuations sanitaires aéroportées de patients adultes de réanimation depuis la Corse
No full textIntroduction: Corsica, an island territory without a university hospital and with limited specialized medical resources, relies heavily on air medical transfers (EVASAN) for optimal patient care. This study aimed to analyze the indications for EVASAN among adult Intensive Care Unit (ICU) patients in order to identify gaps in local medical and surgical capacities.Methods: this retrospective, observational, multicenter study was conducted from January 2022 to December 2024. All adult patients transferred from Corsica to a mainland ICU were included. Data were extracted from Emergency Medical Service (SAMU) Corse-du-Sud/Haute-Corse (2A/2B) intervention forms and medical records. Variables collected included demographics, severity of illness (Simplified Acute Physiology Score II, SAPS II), transfer origin, invasive mechanical ventilation, vasopressor use, procedures performed before transfer, receiving ICU, and reason for EVASAN.Results: among 2,117 EVASANs, 167 (11%) involved ICU patients, 67% of whom were transferred from one of the two Corsican ICUs (the remainder directly from emergency departments). Patients were severely ill (mean SAPS II: 47; intubated: 41%; vasopressors: 26%). The main reasons for transfer were highly specialized surgery (35%), complex multidisciplinary medical management (26%), and interventional radiology (20%). Most patients were transferred to Marseille (75%) or Nice (22%). No in-flight deaths or major complications occurred.Conclusion: in Corsica, EVASAN of ICU patients is frequent, mainly reflecting the absence of highly specialized surgery and interventional radiology. These findings underscore the need to strengthen local expertise and infrastructure in order to reduce dependence on mainland transfers.Introduction : en Corse, territoire insulaire dépourvu de Centre hospitalier et universitaire et à ressources médicales spécialisées incomplètes, l’évacuation sanitaire aérienne (EVASAN) est essentielle pour l’accès aux soins de malades nécessitant un plateau technique élaboré. L’objectif de cette étude est d’analyser les motifs d’EVASAN de patients relevant de réanimation afin d’identifier de possibles solutions au déficit du plateau technique local.Méthodes : étude rétrospective, observationnelle et multicentrique, menée de janvier 2022 à décembre 2024. Ont été inclus les patients adultes relevant de réanimation, transférés depuis la Corse vers un service de réanimation continental. Les données provenaient des fiches du Service d’aide médicale urgente (SAMU) Corse-du-Sud/Haute-Corse (2A/2B) et des dossiers médicaux. Les variables étudiées comprenaient données démographiques, gravité (indice de gravité simplifié, IGS II), origine, support ventilatoire, support par amines durant le transfert, actes avant transfert, service d’accueil et motif d’EVASAN.Résultats : sur 2117 EVASAN, 167 concernaient des patients de réanimation (11 %), dont 67 % provenaient d’une réanimation insulaire (le tiers restant étant orienté directement à partir des services d’urgence). Les patients étaient graves (IGS II moyen 47; 41 % intubés; 26 % sous amines). Les principaux motifs étaient la chirurgie spécialisée complexe (35 %), les prises en charge médicales multidisciplinaires (26 %) et la radiologie interventionnelle (20 %). Les destinations étaient principalement Marseille (75 %) et Nice (22 %). Aucun décès ni complication majeure en vol n’ont été rapportés.Conclusion : en Corse, l’EVASAN de patients de réanimation est fréquente, motivée surtout par une insuffisance du plateau chirurgical et de radiologie interventionnelle. Ces résultats soulignent la nécessité de renforcer certaines compétences et infrastructures afin de limiter ces transferts
Evaluation of geographical distortions in language models
No full textInternational audienceGeographic bias in language models (LMs) is an underexplored dimension of model fairness, despite growing attention being given to other social biases. We investigate whether LMs provide equally accurate representations across all global regions and propose a benchmark of four indicators to detect undertrained and underperforming areas: (i) indirect assessment of geographic training data coverage via tokenizer analysis, (ii) evaluation of basic geographic knowledge, (iii) detection of geographic distortions, and (iv) visualization of performance disparities through maps. Applying this framework to ten widely used encoder-and decoder-based models, we find systematic overrepresentation of Western countries and consistent underrepresentation of several African, Eastern European, and Middle Eastern regions, leading to measurable performance gaps. We further analyse the impact of these biases on downstream tasks, particularly in crisis response, and show that regions most vulnerable to natural disasters are often those with poorer LM coverage. Our findings underscore the need for geographically balanced LMs to ensure equitable and effective global applications
Inverse design of arbitrary large-scale metalenses with a physically-tailored deep learning model
No full textInternational audienceEfficient large-scale metalenses with oblique incidence have drawn a lot of attention for designing future augmented and virtual reality devices. Classical methods based on traditional lookup tables for designing such metalenses lacked efficiency and presented several engineering challenges to achieve high-quality images. This study introduces a novel inverse design methodology to design large-scale metalenses that combines high performance, scalability, and exceptional design flexibility. Our innovative approach assembles multiple optimized beam deflectors that redirect incident light to the focal point. Leveraging deep learning models and incorporating a physics-aware assembling strategy into the optimization process, the high efficiency of the meta-deflectors is translated into high focusing efficiency of the final metalens design. Once the base models are trained, the inverse design process can generate large-scale metalenses with varying sizes, numerical apertures, and focal positions, without the need to re-initialize the entire optimization process, enabling rapid prototyping and versatility in designing optical devices
Fine-Tuning Detection Criteria for Enhancing Anomaly Detection in Time Series
No full textInternational audienceAnomaly detection is the problem of identifying observations that do not conform to the typical ones in a time series. Detection methods implicitly define detection criteria, such as deviation measures, filter thresholds, and candidate anomaly selection strategies. Choosing inappropriate criteria results in inaccurate outputs, generating spurious alerts or missing events. Adjusting these criteria is essential for monitoring systems. To address this challenge, this paper explores the fine-tuning of deviation measures, filter thresholds, and candidate selection strategies. Experimental results show that the proper choice of criteria significantly improves anomaly detection performance, often with greater impact than changing the detection methods
Marker Guess and Check Plus (MGC+): An Efficient Short Blocklength Code for Random Edit Errors
No full textInternational audienceMotivated by DNA-based data storage, we focus on the design of short blocklength codes for correcting random edits, i.e., deletions, insertions, and substitutions. Building on the recently introduced Guess & Check Plus (GC+) code, we propose Marker-GC+ (MGC+), which inserts periodic markers and leverages a marker-aided decoding algorithm to reduce complexity and enhance error correction capability. Our decoder combines trellis-based offset estimation with an informed guessing strategy to achieve these gains. Simulations demonstrate that MGC+ achieves higher code rates, lower decoding times, and improved reliability compared to existing codes
Multi-isotope analysis of mammalian fauna from the Middle Pleistocene (MIS6) Lazaret Cave: palaeoecological, palaeoenvironmental and archaeological implications
No full textInternational audienceCaves and rock shelters, rich in archaeological and palaeoecological evidence, offer exceptional archives for reconstructing past climate and environments, as well as human and animal ecologies of the Pleistocene. In this study, we explore possible changes in dietary and movement ecology of mammalian fauna from the Middle Pleistocene (MIS 6) Lazaret Cave, France, to infer broader palaeoenvironmental conditions of a classic southern glacial refugia. Tooth enamel from various MIS 6 mammalian fauna, including wolves and ungulates, was sampled for carbon (δ13C), oxygen (δ18O) and strontium (87Sr/86Sr) isotopic analysis. δ13C and δ18O data demonstrate consistent differences in isotopic niche space between the carnivore and herbivores, and amongst the different ungulate species, largely consistent with the results of previous dental microwear studies. Strontium isotope data from red deer and ibex demonstrate use of diverse landscapes nearby the site. Their habitat ranges largely overlap, except during Complex CII sup. where a spatial partitioning can be observed. δ 18 O data from red deer was used to reconstruct mean annual temperature (MAT) and results indicate colder climatic values than today, but relatively elevated ranges of estimated temperature compared to other paleoclimate proxies. Together, these findings provide new insights into Middle Pleistocene faunal palaeoecology, local environmental conditions, and hominin activity at Lazaret Cave, within the context of its role as a southern refugium during a glacial period
Closing the Link [Women in Engineering]
No full textInternational audienceA wireless link budget sets the requirements for communication system design, including antennas, for applications ranging from radio astronomy and satellite communication to implantable devices. The ultimate goal of a link budget is to achieve a realistic and acceptable margin to close the link. Achieving a career goal, on the other hand, bears resemblance to constructing of a link budget. This article draws connections between link budget techniques and approaches that might help an early-career antenna engineer close the gap between the personal starting point and the dream goal
Analyse de la distribution spatio-temporelle des accidents technologiques engendrés par des aléas naturels (NaTech) : Une contribution à la connaissance du processus d’alerte en France
No full textInternational audienc
Euclid Quick Data Release (Q1): From spectrograms to spectra: the SIR spectroscopic Processing Function
No full textInternational audienceThe Euclid space mission aims to investigate the nature of dark energy and dark matter by mapping the large-scale structure of the Universe. A key component of Euclid's observational strategy is slitless spectroscopy, conducted using the Near Infrared Spectrometer and Photometer (NISP). This technique enables the acquisition of large-scale spectroscopic data without the need for targeted apertures, allowing precise redshift measurements for millions of galaxies. These data are essential for Euclid's core science objectives, including the study of cosmic acceleration and the evolution of galaxy clustering, as well as enabling many non-cosmological investigations. This study presents the SIR processing function (PF), which is responsible for processing slitless spectroscopic data. The objective is to generate science-grade fully-calibrated one-dimensional spectra, ensuring high-quality spectroscopic data. The processing function relies on a source catalogue generated from photometric data, effectively corrects detector effects, subtracts cross-contaminations, minimizes self-contamination, calibrates wavelength and flux, and produces reliable spectra for later scientific use. The first Quick Data Release (Q1) of Euclid's spectroscopic data provides approximately three million validated spectra for sources observed in the red-grism mode from a selected portion of the Euclid Wide Survey. We find that wavelength accuracy and measured resolving power are within requirements, thanks to the excellent optical quality of the instrument. The SIR PF represents a significant step in processing slitless spectroscopic data for the Euclid mission. As the survey progresses, continued refinements and additional features will enhance its capabilities, supporting high-precision cosmological and astrophysical measurements
Euclid Quick Data Release (Q1): The evolution of the passive-density and morphology-density relations between and
No full textInternational audienceThe extent to which the environment affects galaxy evolution has been under scrutiny by researchers for decades. With the first data from Euclid, we can begin to study a wide range of environments and their effects as a function of redshift, using 63 sq deg of space-based data. In this paper, we present results from the Euclid Q1 Release, where we measure the passive-density and morphology-density relations at -1. We determine if a galaxy is passive using the specific star-formation rate, and we classify the morphologies of galaxies using the Sérsic index n and the u-r colours. We measure the local environmental density of each galaxy using the Nth-nearest neighbour method. We find that at fixed stellar mass, the quenched fraction increases with increasing density up to . This result shows the separability of the effects from the stellar mass and the environment, at least at z<0.75. At z>0.75, we observe weak environmental effects, with most high mass galaxies being quenched independently of environment. Up to , the ETG fraction increases with density at fixed stellar mass, meaning the environment also transforms the morphology of the galaxy independently of stellar mass, at low mass. For high mass galaxies, almost all galaxies are early-types, with low impact from the environment. At z>0.75, the morphology depends mostly on stellar mass, with only low-mass galaxies being affected by the environment. Given that the morphology classifications use u-r colours, these are correlated to the star-formation rate, and as such our morphology results should be taken with caution; future morphology classifications should verify these results. To summarise, we identify the passive-density and morphology-density relations at z<0.75, but at z>0.75 the relations are less strong. At z>0.75, the uncertainties are large, and future Euclid data releases are key to confirm these trends