HAL Portal Cnam (French National Conservatory for Arts and Crafts)
Not a member yet
31010 research outputs found
Sort by
Associations between degree of food processing and all-cause and cause-specific mortality: a multicentre prospective cohort analysis in 9 European countries
No full textInternational audienceUltra-processed food (UPF) consumption has been linked with higher risk of mortality. This multi-centre study investigated associations between food intake by degree of processing, using the Nova classification, and all- cause and cause-specific mortality. Methods This study analyzed data from the European Prospective Investigation into Cancer and Nutrition. All-cause mortality and cause-specific mortality due to cancer, circulatory diseases, digestive diseases, Parkinson's disease, and Alzheimer's disease served as endpoints. Hazard ratios (HRs) and 95% CIs were estimated using multivariable Cox proportional hazards regression models. Substitution analyses were also performed. Findings Overall, 428,728 (71.7% female) participants were included in the analysis and 40,016 deaths were documented after 15.9 years of follow-up. UPFs (in percentage grams per day [g/d]) were positively associated with all- cause mortality (HRs per 1-SD: 1.04; 95% CI: 1.02,1.05), as well as mortality from circulatory diseases (1.09; 95% CI: 1.07,1.12), cerebrovascular disease (1.11; 95% CI: 1.05,1.17), ischemic heart disease (1.10; 95% CI: 1.06,1.15), digestive diseases (1.12; 95% CI: 1.05,1.20), and Parkinson's disease (1.23; 95% CI: 1.06,1.42). No associations were found between UPFs and mortality from cancer or Alzheimer's disease. Replacing processed and UPFs with unprocessed/minimally processed foods was associated with lower mortality risk. Interpretation In this pan-European analysis, higher UPF consumption was associated with greater mortality from circulatory diseases, digestive diseases, and Parkinson's disease. The results support growing evidence that higher consumption of UPFs and lower consumption of unprocessed foods may have a negative impact on health
Recommandations et enjeux concernant la construction de scénarios d'impact du changement climatique en assurance santé et vie
No full textFollowing initial pilot stress test exercises proposed by insurance and banking regulators (see e.g. [ACPR, 2024]), insurance and reinsurance companies face the challenge to design climate change impact scenarios up to time horizons ranging from 2050 to 2100, much beyond the traditional 3 to 5-year view usually provided in their Own Risk and Solvency Assessment. We provide general recommendations for the construction of such scenarios and illustrate them in the context of health and mortality risks of a Metropolitan France insurer.À la suite des premiers exercices pilotes de tests de résistance proposés par les régulateurs de l’assurance et de la banque (voir par ex. [ACPR, 2024]), les compagnies d’assurance et de réassurance sont confrontées au défi de concevoir des scénarios d’impact du changement climatique sur des horizons allant de 2050 à 2100, bien au-delà de la perspective traditionnelle de 3 à 5 ans généralement adoptée dans leur propre évaluation des risques et de la solvabilité (ORSA). Nous formulons des recommandations générales pour la construction de tels scénarios et les illustrons dans le contexte des risques liés à la santé et à la mortalité pour un assureur opérant en France métropolitaine
Improving predictive maintenance: Evaluating the impact of preprocessing and model complexity on the effectiveness of eXplainable Artificial Intelligence methods
No full textInternational audienceDue to their performance in this field, Long-Short-Term Memory Neural Network (LSTM) approaches are often used to predict the remaining useful life (RUL). However, their complexity limits the interpretability of their results. So, eXplainable Artificial Intelligence (XAI) methods are used to understand the relationship between the input data and the predicted RUL. Modeling involves making choices, such as preprocessing strategies or model complexity. Understanding how these modeling choices affect the effectiveness of XAI methods is crucial. This paper investigates the impact of two modeling aspects: preprocessing multivariate time series and model complexity, precisely the number of hidden layers, on the quality of the explanations provided by three XAI post-hoc local agnostic methods (Local Interpretable Model-Agnostic Explanations (LIME), SHapley Additive exPlanations (SHAP), and Learning to eXplain (L2X) in the context of the RUL prediction. The quality of the XAI methods is evaluated using eleven metrics, categorized under five properties based on the definitions of interpretability and explainability. Experiments on the C-MAPSS dataset for aero-engine prognostics demonstrate that SHAP often provides better explanations when optimized preprocessing parameters are used. However, variations in these preprocessing parameters affect the quality of the explanation. Additionally, the results suggest no significant correlation between the complexity of the LSTM model and explanation quality, although changes in the number of layers notably influence the precision of SHAP’s explanations
Développer le bien-être des personnes neuro-atypiques en formation en alternance: proposition d'un modèle théorique et agenda de recherche
No full textInternational audienceThis paper explores ways to support the wellbeing of neurodivergent individuals participating in co-operative education (co-op). The authors propose a theoretical model for supporting neurodivergent student wellbeing in co-op, based on the current understanding of wellbeing in WIL and interventions for neurodivergent individuals at work and in higher education. The paper also identifies methodological considerations in neurodiversity research within the WIL context. It then presents a research agenda identifying critical topics for future WIL research. The expected outcomes and implications for WIL practitioners, organizations, and the WIL community are discussed, highlighting the potential for broad adoption
Influence of diameter and scan strategy on the geometrical, microstructural, and mechanical properties of small Inconel 625 L-PBF struts
No full textInternational audienceThe geometries, microstructures, and mechanical properties of vertically built Inconel 625 Laser Powder Bed Fusion (L-PBF) struts were investigated in this study. The influence of strut size (between 0.2 mm and 2 mm) and scan strategy was more specifically addressed. As-built struts exhibit satisfactory geometry and porosity rates, whatever the strut size and scan strategy. Classical columnar grains oriented parallel to the build direction (BD) were obtained, with a < 001 > // BD fiber texture only for the smaller struts (0.2 mm to 0.5 mm), due to the formation of a unique circular melt pool on the whole strut surface. At a smaller scale, the influence of the build strategy is also visible on solidification cells, whose average diameter decreases for outside-in strategies and larger hatching area ratios. The tensile strengths and hardness values are lower for the smaller diameter (0.3 mm) struts and for the inside-out strategies, suggesting the important role played by a finer sub-grain structure and a smaller crystallographic texture on the strengthening of Inconel 625 struts
La chicorée, un remède pour assainir les sols viticoles
No full textInternational audienceL’utilisation de la bouillie bordelaise contre le mildiou provoque une contamination des sols et un dysfonctionnement des écosystèmes viticoles. Pour limiter ses effets toxiques et dans une perspective d’économie circulaire, le projet Vitalicuivre expérimente la culture de la chicorée qui extrait le cuivre de ces sols pollués pour le recycler en alimentation animale
How to: share and reuse data - challenges and solutions from PrIMAVeRa project
No full textInternational audienceBackground: Data sharing accelerates scientific progress and improves evidence quality. Even thoughjournals and funding institutions require investigators to share data, only a small part of studies madetheir data publicly available upon publication. The procedures necessary to share retrospective data forreuse in secondary data analysis projects can be cumbersome.Objectives: Predicting the Impact of Monoclonal Antibodies & Vaccines on Antimicrobial Resistance is aEuropean research project that aims to develop mathematical models and an epidemiological repositoryto assess the impact of vaccines and monoclonal antibodies on antimicrobial resistance (AMR). Toaccomplish the project aim, Work Package 3 was responsible for gathering historical anonymized indi-vidual patient datasets.Sources: Through a systematic search we have identified 108 eligible studies for data sharing; of whicheight have completed all legal requirements and shared their datasets, with data from four infectioussyndromes and seven resistant pathogens. The AMR data gathered in Predicting the Impact of Mono-clonal Antibodies & Vaccines on Antimicrobial Resistance project are publicly available in EuropeanClinical Research Alliance on Infectious Disease epidemiology network platform (https://epi-net.eu/primavera/about/anonymized-individual-patient-data/)
La notion de vérité à l’épreuve de l’intelligence artificielle
No full textInternational audienceCan we apply traditional notions of truth and error to the artificial production of statements? Drawing on a tripartition proposed by Umberto Eco, the article questions the contemporary relevance of the criteria of truth used to evaluate human statements. It highlights that artificial statements are neither true nor false in the same way as those produced by humans, which challenges our conception of language. The article identifies the specificity and limitations of automatic language generation, focusing on machines' inability to anchor themselves in reality through perception. It concludes by outlining the potential of mechanical enunciation and introduces "humility"—an attitude, contrasted with the common concept of explainability—that we might adopt toward artificial intelligences.Pouvons-nous appliquer les notions traditionnelles de vérité et d’erreur à la production artificielle d’énoncés? A partir d’une tripartition proposée par Umberto Eco, l’article s’interroge sur la pertinence contemporaine des critères de vérité utilisés pour évaluer les énoncés humains. Il s’agit de remarquer que les énoncés artificiels ne sont pas vrais ou faux comme ceux produits par les humains, ce qui remet en question notre conception du langage. L’article identifie la spécificité et les limites de la génération automatique de langage dans l’incapacité des machines de s’ancrer au réel par le biais de la perception. Il conclut en indiquant les potentialités de l’énonciation mécanique et nomme “humilité,” l’attitude, opposée au concept courant d’explicabilité, que l’on pourrait adopter vis-à-vis des intelligences artificielles
Macromolecular and mechanical changes in aged silicones
No full textInternational audienceThree room temperature-vulcanized rubbers were thermally aged under air at 250°C or submitted to aminolysis at room temperature. Ageing was followed by uniaxial tensile testing, sol gel measurements in toluene, and Time Domain Double Quantum NMR characterization. Mechanical tests were exploited to identify the parameters of the Ogden model so as to describe the hyperelastic behavior of PDMS. Sol gel and Double Quantum NMR were used to track and quantify macromolecular changes within the samples. A multiscale correlation between the results of each technique was proposed to highlight the understanding of multiscale analysis in rubber ageing