Portail HAL-PSL
Not a member yet
293219 research outputs found
Sort by
Impact of Announced but Unimplemented Bundled Payment Rules on Hospital Coding Practices in France
No full textBundled Payment for Care Improvement (BPCI) models aim to reduce healthcare costs by improving patient pathways and enhancing care coordination. This study investigates the impact of the announced—but not implemented—BPCI reform in France, which was decided in 2019 but never enacted during the study period (2013–2022). The model proposed adjustments to base payments to account for patient comorbidities. In the French context, it also aimed to reduce reliance on rehabilitation hospitals by incorporating comorbidity data into payment calculations. Despite the absence of actual implementation or financial transfers, significant changes in hospital coding practices were observed in anticipation of the reform. We examine how the announcement of a potential additional payment influenced hospital coding behavior, even without formal policy adoption.To measure this behavioral response, we employ a difference-in-differences (DiD) approach, focusing on two key dimensions: (1) Coding behavior for patients not admitted to inpatient rehabilitation after acute care; (2) Coding behavior for patients admitted to rehabilitation following acute care. Analyzing over one million hip replacement cases, the study reveals a sharp rise in the coding of socio-environmental factors—changes seemingly driven more by anticipated financial incentives than by actual clinical complexity. We find that subjective coding is more prone to overcoding than objective clinical characteristics. Our findings underscore a critical issue: to preserve the integrity of bundled payment models and ensure they truly improve care quality and efficiency, policymakers must urgently reconsider the role of coded variables in reimbursement decisions
Precise relocation of the 14 August 2021 M<sub>w</sub> 7.2 Nippes, Haiti, earthquake sequence using broadband and citizen-hosted short-period seismometers
No full textInternational audienceOn 14 August 2021, the Southern Peninsula of Haiti experienced a Mw 7.2 earthquake, 15 years after the devastating Mw 7.0 event that struck the capital city of Port-au-Prince on 12 January 2010. We use the data from a local temporary broadband seismic network, a national network of low-cost seismometers, and regional seismic networks, together with a probabilistic, global-search, non-linear location method (NLL-SSSTcoherence), to obtain a catalog of 5341 precisely relocated events spanning 20 August 2021 to 6 February 2022, with local magnitudes ranging from 0.5 to 5.6. We compute focal mechanisms for a subset of 73 events through waveform inversion. The catalog can be split into aftershocks directly related to the Nippes earthquake rupture process, and two off-rupture clusters. A first one concerns the Anse-à-Veau-Miragoâne area and corresponds mostly to the aftershock sequence of two Mw 5.3 and 4.9 earthquakes that likely activated a segment of the offshore, south-dipping, Jérémie-Malpasse reverse fault system. A second sequence, offshore Jérémie and clustered close to the offshore trace of that same fault, started immediately after the Nippes mainshock and continued during the entire time interval of the present study. The swarm-like temporal distribution of this sequence, as well as evidence for directional propagation of the epicenters, indicate that it was likely driven by fluid migration. We interpret this seismicity as the result of oblique sub-crustal slip on a south-dipping fault which accounts for oblique convergence between the Gonâve and Caribbean plates in southern Hispaniola. Strain in the crust then partitions between reverse faulting on the Jérémie-Malpasse fault system, strike-slip on the Enriquillo fault, and hybrid faulting in between. Seismic hazard assessment for the region should therefore account for faults other than the Enriquillo fault as potential sources for future earthquakes
Neighborhood Effect and Job Search Behaviors
No full textDo neighbors influence job search behavior? We address this question using a Manski-type model with four search channels. Results show that neighbors' use of a channel affects individuals' own use, particularly for signaling one's job search in the media and using personal or professional networks, as opposed to more conventional methods such as contacting employers or intermediaries. We also find effects of neighbors' occupations. Our findings suggest that local social interactions may amplify labor market inequalities across neighborhoods, as there are stronger incentives to search when unemployed neighbors are actively searching and employed neighbors hold higher-status jobs
Approaching Kasteleyn transition in frustrated quantum Heisenberg antiferromagnets
No full textWe show that the Kasteleyn transition, the abrupt proliferation of infinite strings of defects in classical dimer and related models, can also be relevant for frustrated 2d quantum magnets. This is explicitly demonstrated in a phase of the spin-1/2 Heisenberg diamond-decorated honeycomb lattice where a family of exact eigenstates built as products of dimer and plaquette singlets can be mapped onto the dimer coverings of the honeycomb lattice. The low-temperature properties of this phase are accurately described by an effective dimer model with anisotropic activities and a small, tunable density of monomers, leading to an arbitrarily sharp crossover version of the Kasteleyn transition. The generalization to other geometries and the possibility to realize this model in organo-metallic compounds are briefly discussed
Recommender system in X inadvertently profiles ideological positions of users
No full textStudies on recommendations in social media have mainly analyzed the quality of recommended items (e.g., their diversity or biases) and the impact of recommendation policies (e.g., in comparison with purely chronological policies). We use a data donation program, collecting more than 2.5 million friend recommendations made to 682 volunteers on X over a year, to study instead how real-world recommenders learn, represent and process political and social attributes of users inside the so-called black boxes of AI systems. Using publicly available knowledge on the architecture of the recommender, we inferred the positions of recommended users in its embedding space. Leveraging ideology scaling calibrated with political survey data, we analyzed the political position of users in our study (N=26,509 among volunteers and recommended contacts) among several attributes, including age and gender. Our results show that the platform's recommender system produces a spatial ordering of users that is highly correlated with their Left-Right positions (Pearson rho=0.887, p-value < 0.0001), and that cannot be explained by socio-demographic attributes. These results open new possibilities for studying the interaction between human and AI systems. They also raise important questions linked to the legal definition of algorithmic profiling in data privacy regulation by blurring the line between active and passive profiling. We explore new constrained recommendation methods enabled by our results, limiting the political information in the recommender as a potential tool for privacy compliance capable of preserving recommendation relevance
Universality of the Weyl–Heisenberg symmetry and its covariant quantizations
No full textInternational audienceThe Weyl–Heisenberg symmetries originate from translation invariances of various manifolds viewed as phase spaces, e.g. Euclidean plane, semi-discrete cylinder, torus, in the two-dimensional case, and higher-dimensional generalizations. In this paper, we describe, on an elementary level, how this symmetry emerges through displacement operators and standard Fourier analysis, and how their unitary representations are used both in Signal Analysis (time–frequency techniques, Gabor transform) and in quantum formalism (covariant integral quantizations and semi-classical portraits). An example of application of the formalism to the Majorana stellar constellation in the plane is presented
Risk-taking behavior related to mercury contamination in a high Arctic seabird
No full textInternational audienc
Identification of transdiagnostic phenomena among patients, the general population, relatives, and mental health professionals using topic modeling techniques
No full textInternational audienceIntroduction Recent research has highlighted the limitations of the categorical approach to mental disorders and has increasingly supported the development of a transdiagnostic perspective. This emerging approach focuses on common distal factors (circumstantial, biological, and social) and psychological processes that contribute to psychological suffering across a range of disorders, as well as on the resulting psychological symptoms. The present study aims to identify transdiagnostic distal factors, psychological processes, and symptoms by analyzing narratives through topic modeling—an unsupervised machine learning technique, specifically within Natural Language Processing (NLP). Topic modeling enables the automatic extraction of latent themes from unstructured text, making it possible to identify psychological patterns grounded in patients’ lived experiences. Methods We recruited four groups of participants: Patients diagnosed with a psychiatric disorder ( N = 445), Individuals from the general population ( N = 570), Relatives of patients with psychiatric disorders ( N = 354), and Mental health professionals ( N = 131). Participants answered open-ended questions exploring the causes of psychological suffering, their wishes for change, and their previous experiences with psychotherapy. Results We identified 258 topics, which were organized into 12 overarching themes. The most prominent topics concerned Emotional and Psychological Difficulties , Family and Social Relationships , and Therapeutic Processes . Each theme showed a comparable prevalence across the different diagnostic categories, supporting the transdiagnostic nature of these phenomena. Conclusion Topic modeling can be used effectively to identify transdiagnostic distal factors, psychological processes, and symptoms from diverse narratives. This approach tends to provide a novel means of supporting the relevance and validity of the transdiagnostic perspective
Optimal scaling laws in learning hierarchical multi-index models
No full textIn this work, we provide a sharp theory of scaling laws for two-layer neural networks trained on a class of hierarchical multi-index targets, in a genuinely representation-limited regime. We derive exact information-theoretic scaling laws for subspace recovery and prediction error, revealing how the hierarchical features of the target are sequentially learned through a cascade of phase transitions. We further show that these optimal rates are achieved by a simple, target-agnostic spectral estimator, which can be interpreted as the small learning-rate limit of gradient descent on the first-layer weights. Once an adapted representation is identified, the readout can be learned statistically optimally, using an efficient procedure. As a consequence, we provide a unified and rigorous explanation of scaling laws, plateau phenomena, and spectral structure in shallow neural networks trained on such hierarchical targets
Exposure to a mixture of organic pollutants in a threatened freshwater turtle Emys orbicularis: effects of age, sex, and temporal variation
No full textInternational audienceFreshwater ecosystems constitute major sinks for organic contaminants, increasing anthropogenic pressures and threatening the unique biodiversity they harbour. In addition to persistent legacy compounds, such as polychlorinated biphenyls (PCBs) and organochlorine pesticides (OCPs), various pollutants are less persistent but are chronically released, including polycyclic aromatic hydrocarbons (PAHs), phthalate diesters (PAEs), pyrethroid pesticides, and insect repellent. Heretofore, these pollutants have received insufficient attention in freshwater reptiles, considering their potential to trigger detrimental effects on organisms. During two years (2019 and 2020), we quantified plasma levels of 46 compounds from 7 chemical families in two monitored populations of the protected European pond turtle (Emys orbicularis) in the Camargue wetland, France. PAHs and PAEs were found predominantly and concomitantly, with similar occurrences and levels in the two populations. We observed similar inter-annual variations in PAHs and PAEs with differences between males and females, highlighting the need for a better assessment of the role of sex in the exposure pathway and the toxicokinetics of contaminants, especially in turtles. The negative relationship between PAH levels and age, as well as the high intra-individual variation in levels of both contaminant families, provides further evidence of limited bioaccumulation of these pollutants in the blood of E. orbicularis. This could be explained by the metabolic biotransformation of parent compounds, highlighting the need to quantify the levels of PAH metabolites and phthalate monoesters. Finally, our work underscores the importance of long-term monitoring to better determine the vulnerability of turtle populations already exposed to a wide range of contaminants.</div