BACKGROUND:The aim of the study was to assess overall and cause-specific mortality among employees and former employees of the main public transport company in the Paris region. METHODS:The cohort included all personnel employed between 1980 and 2012 with at least 1 year of service. Vital status and causes of death were obtained from national registries and classified according to the European shortlist for causes of death. Occupations were categorized into 22 distinct groups, and complete occupational histories were reconstructed. Cox proportional hazards models were used to estimate survival by socio-professional group and hazard ratios of death by occupation. Standardized mortality ratios (SMRs) were calculated to compare mortality with that of the general population of the Paris region. RESULTS:A total of 96,634 individuals (78,702 men and 17,932 women) were included. By the end of follow-up, 14.6% of men and 6.5% of women had died. A social gradient in survival was observed with marked differences in the hazard ratios of mortality across occupational categories. Among men, significant excess mortality was found for kidney cancer (SMR = 1.24), acute myocardial infarction (SMR = 1.17), chronic liver disease (SMR = 1.11), and suicide (SMR = 1.20). Among women, mortality due to transport accidents was significantly in excess (SMR = 1.85). Specific excesses were found by occupation: for example, in men, tumors in skilled maintenance workers (SMR = 1.67), ischemic heart disease in bus drivers (SMR = 1.21), security staff (SMR = 2.02), and unskilled workers (SMR = 1.18), and in women, cerebrovascular diseases in bus drivers (SMR = 1.85). CONCLUSION:These findings highlight occupational inequalities in mortality and support strengthening targeted prevention strategies for high-risk workgroups.
Automated analysis of customer feedback on social media is hindered by three challenges: the high cost of annotated training data, the scarcity of evaluation sets, especially in multilingual settings, and privacy concerns that prevent data sharing and reproducibility. We address these issues by developing a generalizable synthetic data generation pipeline applied to a case study on customer distress detection in French public transportation. Our approach utilizes backtranslation with fine-tuned models to generate 1.7 million synthetic tweets from a small seed corpus, complemented by synthetic reasoning traces. We train 600M-parameter reasoners with English and French reasoning that achieve 77-79
Contexte Au cours de la dernière décennie, la réglementation française en matière de santé au travail a connu une transformation majeure, passant d’une attention centrée sur l’aptitude individuelle à une approche globale et préventive. La loi no 2016-1088 a amorcé ce changement, suivie par la loi no 2021-1018 qui a élargi les missions des services de santé au travail, devenus services de prévention et de santé au travail (SPST). Ces évolutions traduisent une volonté de placer la prévention et la qualité de vie au travail au cœur des politiques de santé.Dans ce contexte, le SPST de la RATP a initié, en partenariat avec Santé Publique France, la mise en place d’une étude épidémiologique pour suivre les trajectoires des agents et ex-agents de la RATP en matière de santé (STARS). Méthode La cohorte STARS utilise les auto-questionnaires proposés par CONSTANCES, grande cohorte nationale en population générale, pour décrire les caractéristiques sociodémographiques, le mode de vie, la santé et les expositions professionnelles vie entière, des volontaires à l’inclusion. Un suivi longitudinal sera assuré, via les réponses annuelles aux questionnaires mode de vie et santé, pendant et après un emploi ou une carrière à la RATP. Le protocole soumis à un comité de protection des personnes a reçu un avis favorable le 06 novembre 2025. Le traitement des données personnelles, est mis en œuvre en conformité avec la méthodologie de référence (MR003) de la commission nationale de l’informatique et des libertés (CNIL). Le recueil et les analyses sont réalisées par un tiers de confiance (Cohortia), les données individuelles ne sont accessibles à aucun autre acteur y compris le SPST de la RATP.Dans un second temps, pour les volontaires qui l’acceptent, les données recueillies seront enrichies avec des données provenant du SPST, du laboratoire d’essais mesures ou des ressources humaines. Un chaînage avec le système national des données de santé sera réalisé pour décrire le recours aux soins des volontaires et les causes de décès. Le comité éthique et scientifique pour les recherches, les études et les évaluations dans le domaine de la santé (CESREES) et la CNIL seront alors sollicités pour avis réglementaire.Les données de CONSTANCES seront utilisées comme bras de référence pour limiter les biais d’analyse, avec appariement possible pour des facteurs de confusion tel que la consommation de tabac. Perspectives Les données de la cohorte STARS offriront la possibilité, selon les axes d’intérêt des parties concernées, d’étudier un large éventail de questions sanitaires. Celles-ci permettront, sur des bases scientifiques solides, une orientation et une hiérarchisation des mesures de prévention adaptées aux salariés de l’entreprise.
T-stubs and L-stubs loaded in tension are generally assumed to be in contact with a rigid foundation due to symmetry that influences the prying force and stiffness of these components. However, for double web angle riveted connections the angle rests on deformable beam/column web. The objective of the present paper is to take into account the effects of this deformability on the development of the contact area, the rivet forces and the tensile stiffness. For this purpose, an enhanced theory of beam in contact with a tensionless Winkler foundation is adopted in the contact and lift-off areas and the axial/bending flexibility of the rivet are also considered. The two legs of the angle are fully modelled in interaction with rivets. A numerical analysis is also developed with ANSYS in order to conduct a sensitivity study to assess the influence of the horizontal leg thickness. The analytical model determines with a good accuracy the rivet force, initial stiffness, position of the prying forces but also the shape of the contact pressure distribution.
This paper presents a discrete-event model for a mass-transit line operated with a two-service skip-stop policy while allowing for train dwell times to vary according to passengers’ demand volumes. The model is formulated by two mathematical constraints on the train’s travel and safe separation times that govern the train dynamics on the line. In addition, the model takes into account trains’ dwell times, which are affected by both the services offered by the operator and passenger demand. The model is written in the max-plus algebra, a mathematical framework that allows us to derive interesting analytical results, including the fundamental diagram of the line, which describes the relationship between the average train time headway (or frequency), the number of trains running on the line and the passenger travel demand. The paper also derives indicators that are capable of quantifying and, thus, assessing the impact of a skip-stop policy on passengers’ travel. Finally, the paper compares two different passenger demand profiles. Results show that long-distance passengers mainly benefit from skip-stop policies, while short-distance travelers may experience an increase in their travel time. For long-distance passengers, the increase in the waiting time is counterbalanced by the decrease in the in-vehicle time, leading to an overall decrease in total passenger travel time.
