HAL Portal Univ-Côtedazur
Not a member yet
95391 research outputs found
Sort by
Efficient and Robust Transfer Learning of Optimal Individualized Treatment Regimes with Right-Censored Survival Data
No full textInternational audienceAn individualized treatment regime (ITR) is a decision rule that assigns treatments based on patients' characteristics. The value function of an ITR is the expected outcome in a counterfactual world had this ITR been implemented. Recently, there has been increasing interest in combining heterogeneous data sources, such as leveraging the complementary features of randomized controlled trial (RCT) data and a large observational study (OS). Usually, a covariate shift exists between the source and target population, rendering the source-optimal ITR not optimal for the target population. We present an efficient and robust transfer learning framework for estimating the optimal ITR with right-censored survival data that generalizes well to the target population. The value function accommodates a broad class of functionals of survival distributions, including survival probabilities and restrictive mean survival times (RMSTs). We propose a doubly robust estimator of the value function, and the optimal ITR is learned by maximizing the value function within a pre-specified class of ITRs. We establish the cubic rate of convergence for the estimated parameter indexing the optimal ITR, and show that the proposed optimal value estimator is consistent and asymptotically normal even with flexible machine learning methods for nuisance parameter estimation. We evaluate the empirical performance of the proposed method by simulation studies and a real data application of sodium bicarbonate therapy for patients with severe metabolic acidaemia in the intensive care unit (ICU), combining a RCT and an observational study with heterogeneity.</div
Select and Augment Segmented Items (SASI) : une approche d'agrandissement par objet pour la reconnaissance dynamique des visages chez les personnes atteintes d'une perte de vision centrale
No full textSupporting social interactions is a major challenge for people with central vision loss, especially when it involves recognizing moving faces. In this study, we compare different magnification strategies in a dynamic face recognition task designed to emulate a challenging scenario for low vision patients. Among these strategies, we introduce an item-based magnification approach, Select and Augment Segmented Items (SASI), based on interaction principles previously explored only in static 2D contexts. The core idea of SASI is to allow patients to select specific segmented items (here, faces) and stabilize their magnified version in space, enabling easier and more accurate analysis. Following these principles, we introduce two SASI-based variants adapted to dynamic 3D environments. Using Virtual Reality as a controlled yet ecologically valid platform, we show that SASI magnifiers offer clear advantages over traditional head-centered magnification, leading to more accurate face recognition. This work is the first to demonstrate the effectiveness of SASI principles in complex, dynamic scenarios, highlighting their potential to enhance visual support for social interaction
De nouvelles architectures pour les Big Data
No full textEAN : 9782271153739International audienceContrairement aux systèmes SQL, conçus pour gérer des données structurées, les environnements big data reposent sur une architecture logicielle organisée en trois couches indépendantes : le stockage, la gestion des données et l'analyse. Cette sépara/on des couches a favorisé l'émergence d'un écosystème riche en ou/ls, frameworks et systèmes interopérables, fondés sur des architectures distribuées et parallèles. Dans ce chapitre, nous reviendrons sur la définition du big data ainsi que sur la gestion distribuée et parallèle des données, afin de présenter ces nouvelles architectures et leur intégration dans un environnement cloud
Class-Specific Variable Speed Limit for Traffic Flow Optimization on Road Networks
No full textInternational audienceThis paper presents a novel approach to traffic management in road networks, consisting in time-varying classspecific variable speed limit (VSL) restricted to a fraction of road users. In particular, we present a macroscopic approach where traffic dynamics is described by a multi-class Lighthill-Whitham-Richards (LWR) model, with two classes of users (controlled and uncontrolled vehicles). The model can be applied to a general road network and our goal is to optimize traffic performance by minimizing both the average travel time in the network and the total time spent in virtual buffers at the network entries to prevent spill-back scenarios. The optimization is performed by acting on different ratios of controlled vehicles, and we compare the performance of the proposed control strategy with a classical inflow control at the entries of the network. The numerical tests show that class-specific speed control outperforms inflow control, and highlight the importance of tailored traffic control strategies for road networks, offering insights into optimizing mobility, safety, and traffic efficiency
Reliable Multi-Level Optimization for Safe Predictive Control of Autonomous Vehicles to Avoid Uncertain Multimodal PLEVs
No full textInternational audienceSafety assurance using all perceptual information to predict the motion of dynamic agents is critical in urban environments and remains an open challenge. For Autonomous Vehicles (AV) operating around vulnerable road users, the risk assessment strategy often needs to address stochastic uncertainties in the multiple possible trajectories (or multimodal motion) of the surrounding traffic agents. However, this increases the complexity of the navigation problem using the existing planners. To address this issue, this paper presents a multilevel optimization strategy that combines sampling-based and direct optimization methods for decision-making and control with improved safety and trajectory smoothness. In the primary stage, a sampling-based optimization framework systematically identifies safe candidate trajectories by employing the Fusion of stochastic Predictive Inter-Distance Profile (F-sPIDP). F-sPIDP encapsulates the multimodal dynamics of traffic agents and explicitly computes the uncertainties in their estimated or tracked states. From the set of trajectories, a reference optimal trajectory and its F-sPIDP setpoints are selected, adhering to stringent safety constraints and motion smoothness. Subsequently, a secondary local control optimization refines the optimal trajectory to ensure compliance with the AV's kinematic and dynamic constraints while accounting for the quantified uncertainty within the F-sPIDP framework. The performance of the proposed method was assessed through simulations and statistical analyses, evaluating its robustness to diverse levels of uncertainty.</div
Mechanistic Modeling of Joint Circulating Cell-free DNA Concentration—Tumor Size Kinetics under Immune-Checkpoint Inhibitors in Advanced Cancer
No full textInternational audienceObjectivesIn cancer patients, the plasmatic concentration of global cfDNA — shed by both tumor and wild-type (especially immune) cells — rises and fluctuates with tumor size, number of metastases, and treatment[1]. Analyzing cfDNA concentration and fragment sizes[2] (part of the fragmentome) is the topic of intense current research and part of the emerging field of liquid biopsy. Given the unclear biology underlying cfDNA release, fragmentation and elimination, mechanistic modeling of joint tumor-cfDNA longitudinal data can shed light on the relationship between tumor kinetics (TK), cfDNA levels, and treatment outcome. The ongoing SChISM (Size cfDNA Immunotherapies Signature Monitoring) clinical study monitors plasma cfDNA size profiles during immune-checkpoint inhibition (ICI). It aims to better understand cfDNA biology during treatment and enable early therapeutic adjustments to mitigate ICI-related progression or toxicity.The objectives of the present work were to: 1) Develop a mechanistic model of cfDNA–TK from clinical data of cancer patients undergoing ICI.2) Study the association of early, on-treatment, model-based parameters with clinical outcome.Methods DataOne hundred and thirty-nine advanced cancer patients treated with ICI, either as monotherapy or in combination therapy, were enrolled. Cancer types comprised non-small cell lung cancer (NSCLC, N = 60), squamous cell carcinomas of the head and neck (N = 28), metastatic clear cell renal cancer (N = 13), advanced urothelial bladder carcinomas (N = 9), and melanoma (N = 29). Using the patented cost-effective BIABooster[3,4] device from Adelis Technologies (France), cfDNA size profiles were generated from plasma samples collected at baseline and at each drug administration. The primary endpoint was early progression (EP), defined as the progression at first imaging, around 3 months after treatment initiation. The secondary endpoint was progression-free survival (PFS).ModelA mechanistic model was developed to jointly describe the dynamics of global cfDNA concentration and tumor lesions under immunotherapy. The model relied on the following biological assumptions: - tumor cells (assimilated to the sum of largest diameters S(t)) comprise two sub-populations: cells sensitive to treatment Ss and resistant ones Sr, governed by first ordered kinetics driven by a shrinkage parameter KS and a regrowth parameter KG.- cancer cells induce the release of DNA fragments (concentration C(t)) due to their chaotic apoptosis-proliferation cycle characteristics[5], with a rate λ.- cfDNA is eliminated from the circulation by a process depending only on its concentration f(C).dSs / dt = KS . SsdSr / dt = KG . SrS = Sr + SsdC/dt = λ . S - f(C)with initial conditions Ss(t=0) = Ss_0, Sr(t=0) = Sr_0 and C(t=0) = C_0, where t=0 denotes treatment initiation.We compared three hypotheses for cfDNA clearance by the liver and kidneys[6]:f_1 (C) = k_D : constant clearance.f_2 (C) = k_D . C : linear (proportional) clearance.f_3 (C) = k_D . C / (C_50 + C) saturating clearance.Mixed-effects modeling To account for inter-individual variability, random effects were assumed to have log-normal distribution on the individual parameters. The best error models were constant for the TK estimation and proportional for cfDNA one, which minimized the corrected Bayesian information criterion. The population parameters were identified using the Monolix software and the Stochastic-Approximation Expectation Maximization algorithm[7]. Tumor kinetics (TK) population parameters (Ss_0 , Sr_0, KS, and KG) were identified from radiological report data and then fixed in the joint cfDNA-TK model. In a second step for association with PFS, the individual empirical Bayes estimates were re-identified using longitudinal data truncated at 1.5 months, using 118 patients who had not progressed. The population parameters prior was fixed from the full kinetics (FK) model for Bayesian estimation.Association with clinical outcomeThe truncated empirical Bayes parameter estimates were pooled with pre-treatment fragmentome-derived metrics (e.g., concentration within multiple size ranges, position of the distribution peaks) to assess associations with the endpoints, using univariable and multivariable logistic/Cox proportional hazard regression models. ResultsThe double exponential model with constant error provided the best fit for TK. The final TK-cfDNA model assumed proportional clearance f_2 (C) and treated monotherapy and combination therapy equivalently. It was able to accurately describe non-trivial cfDNA kinetics. Notably, it captured both steady trends in cfDNA concentration and transient spikes at treatment initiation. These observations were impossible to generate under models with constant or saturated clearance, indicating that proportional cfDNA clearance is a likely mechanism.Goodness-of-fit diagnostics indicated no model misspecification: individual conditionally weighted residuals were centered around zero, all population parameter relative standard errors were below 19%, and parameter estimates absolute correlations did not exceed 0.22. A condition number of 3.13 further confirmed the model was not over-parameterized.In statistical FK analyses, higher KG was significantly associated with EP and shorter PFS, as well as Sr_0. Both outperformed the best baseline cfDNA metric, i.e., the relative concentration of long fragments (≥ 1650 base pairs). CfDNA FK parameters (C_0 and k_D) were significantly associated with PFS only for NSCLC patients (log-rank test: p-value = 0.007 and 0.03, respectively).These results remained consistent for KG in the truncated analysis (OR UV: 1.66 [CI: 1.14-2.42], p = 0.008; AUC: 0.74; HR UV: 1.49 [CI: 1.19-1.86], p < 0.001; C-index: 0.67). KG also outperformed the best baseline cfDNA metrics, i.e., long fragments (≥ 1650 base pairs), conversely to Sr_0. KS was also associated with longer PFS (HR: 0.31 [CI: 0.15-0.64], p = 0.003; C-index: 0.8), as well as the proportion of resistant cells at baseline (HR: 1.37 [CI: 1.05-1.77], p = 0.02; C-index: 0.63). ConclusionsOur mechanistic modeling approach supports a proportional clearance of cfDNA and successfully captured cfDNA “bumps” at treatment initiation. Consistent with previous findings, early tumor kinetics parameters (growth and decay rates, as well as the baseline proportion of resistant cells) were associated with outcome. The parameters specific to cfDNA kinetics correlated with PFS only in the NSCLC subgroup and when considering FK.Future work will explore fragment size-dependent release and fragmentation mechanisms, to account for different biological mechanisms of cfDNA shedding (e.g., apoptosis for short fragments and necrosis for longer fragments[6]). Integrating joint modeling with survival outcomes is also an interesting avenue, alongside with multivariable analyses leveraging machine learning techniques.References1.Leon SA, Shapiro B, Sklaroff DM, Yaros MJ. Free DNA in the serum of cancer patients and the effect of therapy. Cancer Res. 1977;37(3):646-650.2.Qi T, Pan M, Shi H, Wang L, Bai Y, Ge Q. Cell-Free DNA Fragmentomics: The Novel Promising Biomarker. Int J Mol Sci. 2023;24(2):1503. doi:10.3390/ijms240215033.Boutonnet A, Pradines A, Mano M, et al. Size and Concentration of Cell-Free DNA Measured Directly from Blood Plasma, without Prior DNA Extraction. Anal Chem. 2023;95(24):9263-9270. doi:10.1021/acs.analchem.3c009984.Andriamanampisoa CL, Bancaud A, Boutonnet-Rodat A, et al. BIABooster: Online DNA Concentration and Size Profiling with a Limit of Detection of 10 fg/μL and Application to High-Sensitivity Characterization of Circulating Cell-Free DNA. Anal Chem. 2018;90(6):3766-3774. doi:10.1021/acs.analchem.7b040345.Heitzer E, Auinger L, Speicher MR. Cell-Free DNA and Apoptosis: How Dead Cells Inform About the Living. Trends in Molecular Medicine. 2020;26(5):519-528. doi:10.1016/j.molmed.2020.01.0126.Kustanovich A, Schwartz R, Peretz T, Grinshpun A. Life and death of circulating cell-free DNA. Cancer Biol Ther. 2019;20(8):1057-1067. doi:10.1080/15384047.2019.15987597.Kuhn E, Lavielle M. Maximum likelihood estimation in nonlinear mixed effects models. Computational Statistics & Data Analysis. 2005;49(4):1020-1038. doi:10.1016/j.csda.2004.07.00
The Triangle of Misunderstanding in Interactive Virtual Narratives: Gulfs Between System, Designers and Players
No full textInternational audienceDesigners of storytelling experiences in virtual reality (VR) can take advantage of the medium's realism and immersion to communicate their intentions. However, interaction freedom comes with unpredictability, raising the risk of miscommunication between the experience sought by the designer and the player's interpretation. To better understand such miscommunications, we revisit Don Norman's work on stages of action to propose a model of designerplayer gulfs in VR that incorporates eight classes of communication gulfs. We designed a two-phase study where 10 participants designed VR scenarios and then played scenarios created by previous participants. Through coupled structured interviews, we identified 127 issues in VR-mediated communication that were mapped to our model to understand their impact on the player's interpretation of the narrative experience. Our work provides a roadmap to identifying sources of miscommunication in VR, a first step to conceiving principles and guidelines for achieving effective communication in storytelling experiences
Cell-Free DNA Fragmentome Dynamics as a Biomarker for Immune-Checkpoint Inhibition in Advanced Carcinoma
No full textInternational audienceEarly prediction of resistance to immunotherapy is a major challenge in oncology. The ongoing SChISM (Size Cell-fre DNA (cfDNA) Immunotherapies Signature Monitoring) clinical study proposes an innovative approach based on patented cfDNA quantification methods, providing concentration and size profile fluctuations of plasmatic circulating DNA for early therapeutic management of immune checkpoint inhibitors treated patients. The main interest is that such measures can be performed much earlier than the first imaging evaluation, thanks to liquid biopsies. Five cancer types were investigated: melanoma, head and neck, renal, bladder and lung cancers, with a total of 260 patients at the end of the study, described by their clinical and classical biological data, and cfDNA features, as concentration, first and second peak of the cfDNA size distribution, and specific size ranges of cfDNA fragments. The principal purpose is to early predict response to immunotherapy by searching for a longitudinal signature in the concentration and sizes in cfDNA. The study aims to develop mechanistic model of cfDNA joint kinetics with other longitudinal markers and tumor size imaging. Such models embedded within a statistical framework will then be calibrated on our population data. Machine learning models will be used to predict some outcomes as radiologically confirmed progression at the first imaging evaluation, progression-free survival or the overall survival, thanks to these dynamic parameters and other variables available at baseline and in order to predict response to immunotherapy. Thus, typical classification models as logistic regression, or survival models as proportional hazard Cox regression model will be tested to analyze feature at baseline, and models consisting of a dynamic system of differential equations will help us to describe the evolution of the quantitative profile of cfDNA over time
Mechanistic modeling of tumor and size-dependent cell-free DNA kinetics under immune checkpoint inhibition
No full textBackgroundIn cancer patients, the plasmatic concentration and size of cell-free DNA (cfDNA) — shed by both tumor and hematopoietic cells — rise and fluctuate with tumor and immune system development under treatment. Quantified through liquid biopsies, cfDNA has shown promise as a biomarker for treatment monitoring, offering a less invasive and more cost-effective alternative to traditional tissue biopsies. An active growing field, fragmentomics, focuses on cfDNA fragmentation patterns, especially fragment size distributions. However, key mechanisms remain poorly understood, including the link between tumor burden and cfDNA production and fragmentation, the elimination kinetics, and the interplay between cfDNA dynamics and disease progression. To address these gaps, we developed a mathematical mechanistic model that jointly describes the dynamics of short cfDNA fragments (75-580 base pairs (bp)), long cfDNA fragments (580-1650 bp), and tumor size kinetics (TK) under immunotherapy. MethodsTumor cells were assumed to comprise treatment-sensitive T_S and treatment-resistant T_R cells, and to be governed by first ordered kinetics driven by a shrinkage parameter β and a regrowth parameter α. Short and long fragments were assumed to be released through (i) active secretion during proliferation and (ii) apoptosis of T_S, although at different rates. Constant, linear, and saturating cfDNA elimination processes were compared.The model was developed using a cohort of 112 advanced cancer patients treated with immune checkpoint inhibitors, with multiple tumor types. The median number of cfDNA measurements was 7 versus 2 for tumor sizes (sum of the largest diameters). Statistical evaluation of the model against the data was performed using non-linear mixed effects modeling. To assess the ability of the model to provide early on-treatment metrics able to predict clinical outcome, individual model coefficients were re-identified using only the first 6 weeks of data from 90 patients who had not progressed during this period.ResultsThe size-dependent TK-cfDNA model successfully described non-trivial cfDNA patterns, capturing steady trends and transient spikes at treatment initiation, observed in short and/or long fragment dynamics. We then demonstrated that an early model-derived parameter computed at 6 weeks of treatment (ratio of release rate to elimination rate of the short fragments), was significantly associated with post-6 weeks-PFS (HR = 1.3 [95 % confidence interval: 1-1.7], p = 0.03).ConclusionOur results provide the first mathematical model of joint TK and size-dependent cfDNA fragments dynamics validated against clinical data. It not only provides mechanistic insights into the biology of cfDNA tumor release and systemic elimination but also predicts outcome from early on-treatment data, enabling possible therapeutic adjustments to mitigate ICI-related progression or toxicity
The MeerKAT Massive Distant Clusters Survey: Detection of Diffuse Radio Emission in Galaxy Clusters at z > 1
No full textInternational audienceDiffuse, low surface-brightness radio emission in merging galaxy clusters provides insights into cosmic structure formation, the growth of magnetic fields, and turbulence. This paper reports a search for diffuse radio emission in a pilot sample of six high-redshift (1.01 < z < 1.31) galaxy clusters from the MeerKAT Massive Distant Cluster Survey (MMDCS). These six clusters are selected as the most massive systems based on their Sunyaev-Zel'dovich mass from the full MMDCS sample of 30 ACT DR5 clusters, and were observed first to explore the high-mass, high-redshift regime. Diffuse radio emission is confidently detected in four clusters and tentatively identified in two, with -corrected radio powers scaled to 1.4 GHz ranging from to and linear sizes between 0.47 and 1.08 Mpc. Combining X-ray data with MeerKAT radio data, we find that 80 of clusters with X-ray observations exhibit disturbed morphologies indicative of mergers. These z > 1 galaxy clusters scatter around the established radio power-mass scaling relation observed at lower redshifts, supporting turbulent re-acceleration models in high-redshift mergers. However, their radio spectra are predicted to steepen (\alpha < -1.5) due to enhanced inverse Compton losses in the cosmic microwave background, rendering them under-luminous at 1.4 GHz and placing them below the correlation. Our results demonstrate that merger-driven turbulence can sustain radio halos even at z > 1 while highlighting MeerKAT's unique ability to probe non-thermal processes in the early universe