HAL - Lille 3
Not a member yet
5967 research outputs found
Sort by
Agreement among physiotherapists in assessing patient performance of exercises for low-back pain
No full textInternational audienceBACKGROUND: There is no agreement for the performance assessment of patients who practice exercises.. (2 points to withdraw) This assessment is currently left to the physiotherapist's personal judgement. We studied the agreement among physiotherapists in rating patient performance during exercises recommended for chronic low-back pain (LBP).METHODS: A vignette-based method was used. We first identified ten exercises recommended for LBP in the literature. Then, 42 patients with chronic LBP participating in a rehabilitation program were videotaped during their performance of one of the ten exercises. A vignette was an exercise video preceded by clinical information. Ten physiotherapists from primary (4) and tertiary care (6) viewed the 42 vignettes twice, one month apart, and rated patient performance from zero (worse performance) to ten (excellent performance) by considering the position and duration of the contraction or stretching. Intra-class correlation coefficients (ICCs) and 95% confidence intervals (95% CIs) were computed to assess inter- and intra-rater reliability.RESULTS: The overall inter-rater agreement was fair (ICC 0.48 [95% CI 0.33-0.56]) but was better for stretching exercises (0.55 [0.35-0.64]) than strengthening exercises (0.42 [0.20-0.52]) and for tertiary-care physiotherapists (0.66 [0.54-0.76]) than primary-care physiotherapists (0.28 [0.09-0.37]). The intra-rater agreement was overall good (0.72 [0.57-0.81] to 0.88 [0.79-0.94]). It was better for stretching exercises (from 0.68 [0.46-0.81] to 0.96 [0.91-0.98]) than strengthening exercises (from 0.68 [0.38-0.84]) to 0.82 [0.56-0.92]).CONCLUSION: The agreement in rating patient performance of exercises for LBP is good among physiotherapists trained in managing LBP but is low among non-trained physiotherapists
Transfer Learning with deep Convolutional Neural Network for Underwater Live Fish Recognition
No full textInternational audienceRecently, marine ecologists are more interested in using underwater video analysis to study fish populations as this technique is non-destructively, produces large amount of visual data and does not affect fish behavior. Automated methods for processing and analyzing the recorded data are required because visual analysis can be subjective, time consuming and costly. However, the underwater environment poses great challenges due to changes in luminosity, complex backgrounds and free movement of fish. In this paper, we present a convolutional neural network that was trained with transfer learning framework for fish species classification. First, we extract fish features from images using the original AlexNet model on the available underwater dataset. Then, to improve the performance, we fine-tune the model on the dataset. Finally, we re-extract features after that AlexNet has been fine-tuned. For classification, we use a linear SVM classifier. The experiment results demonstrate the effectiveness of the proposed approach, we achieve an accuracy of more than 99%
DO MULTIPLE ADMINISTRATIONS OF STABLE IODINE PROTECT POPULATION CHRONICALLY EXPOSED TO RADIOACTIVE IODINE: WHAT IS PRIODAC RESEARCH PROGRAM (2014-22) TEACHING US?
No full text15th Coordination and Planning Meeting of the World-Health-Organization's Radiation-Emergency-Medical-Preparedness-and-Assistance-Network (WHO-REMPAN), Geneva, SWITZERLAND, JUL 03-05, 2017International audienceSingle dose of potassium iodide (KI) is recommended to prevent the risk of thyroid cancer during nuclear accidents. However in the case of repeated/protracted radioiodine release, a unique dose of KI may not protect efficiently the thyroid against the risk of further developing a radiation-induced cancer. The new WHO guidelines for the use in planning for and responding to radiological and nuclear emergencies identify the need of more data on this subject as one of the four research priorities. The aims of the PRIODAC project are (1) to assess the associated side effects of repeated intakes of KI, (2) to better understand the molecular mechanisms regulating the metabolism of iodine, (3) to revise the regulatory French marketing authorization of 65-mg KI tablets and (4) to develop new recommendations related to the administration of KI toward a better international harmonization. A review of the literature and the preliminary data are presented here
The Mesozoic Along-Strike Tectonometamorphic Segmentation of Longmen Shan (Eastern Tibetan Plateau)
No full textInternational audienceThe Longmen Shan belt (eastern border of the Tibetan plateau) constitutes a tectonically active region as demonstrated by the occurrence of the unexpected 2008 M w 7.9 Wenchuan and 2013 M w 6.6 Lushan earthquakes in the central and southern parts of the belt, respectively. These events revealed the necessity of a better understanding of the long-term geological evolution of the belt and its effect on the present dynamics and crustal structure. New structural and thermobarometric data offer a comprehensive data set of the paleotemperatures across the belt and P-T estimates for low-grade metamorphic domains. In the central Longmen Shan, two metamorphic jumps of 150-200°C, 5-6 kbar and~50°C, 3-5 kbar acquired during the Early Mesozoic are observed across the Wenchuan and Beichuan faults, respectively, attesting to their thrusting movement and unrevealing a major decollement between the allochtonous Songpan-Garze metasedimentary cover (at T > 500°C) and the autochtonous units and the basement (T < 400°C). In the southern Longmen Shan, the only greenschist facies metamorphism is observed both in the basement (360 ± 30°C, 6 ± 2 kbar) and in the metasedimentary cover (350 ± 30°C, 3 ± 1 kbar). Peak conditions were reached at ca. 80-60 Ma in the basement and ca. 55-33 Ma in the cover, ca. 50 Ma after the greenschist facies metamorphic overprint observed in the central Longmen Shan (ca. 150-120 Ma). This along-strike metamorphic segmentation coincides well with the present fault segmentation and reveals that the central and southern Longmen Shan experienced different tectonometamorphic histories since the Mesozoic
Histoires d’espaces. Territoires et limites. Autour de Daniel Nordman
No full textInternational audienceDe l’histoire de la Méditerranée à celle de la géographie, du Maghreb à la France, par les voyages, le déplacement, les frontières, Daniel Nordman aura exploré dans toutes ses dimensions l’espace, objet central de cet ouvrage. Il invite à penser le rapport entre des acteurs et des lieux, des savants et des arrière-pays, des soldats et des cartes. La construction des territoires, l’élaboration de leurs limites, les interactions de part et d’autre de la Méditerranée sont autant d’histoires d’espace que ce livre aborde à travers les contributions de chercheurs qui tous ont été marqués, à un moment ou un autre de leur trajectoire, par les intuitions et les conseils de Daniel Nordman, qui fut leur collègue, leur directeur de thèse, leur enseignant, leur ami. Ces remerciements académiques disent, par la diversité des approches et des objets, l’importance d’une oeuvre et de son héritage intellectuel
Personalized and Private Peer-to-Peer Machine Learning
No full textInternational audienceThe rise of connected personal devices together with privacy concerns call for machine learning algorithms capable of leveraging the data of a large number of agents to learn personalized models under strong privacy requirements. In this paper, we introduce an efficient algorithm to address the above problem in a fully decentralized (peer-to-peer) and asynchronous fashion, with provable convergence rate. We show how to make the algorithm differentially private to protect against the disclosure of information about the personal datasets, and formally analyze the trade-off between utility and privacy. Our experiments show that our approach dramatically outperforms previous work in the non-private case, and that under privacy constraints, we can significantly improve over models learned in isolation
Robust observer-based stabilization of linear discrete-time systems with parameter uncertainties
No full textInternational audienceThis paper addresses the problem of observer-based stabilization of discrete-time linear systems in presence of parameter uncertainties and -bounded disturbances. We propose a new variant of the classical two-steps LMI approach. In the first step, we use a slack variable technique to solve the optimization problem resulting from the stabilization problem by a static state feedback. In the second step, a part of the slack variable obtained is incorporated in the observer-based stabilization problem, to calculate simultaneously the Lyapunov matrix and the observer-based controller gains. A numerical evaluation is presented to show the superiority of the proposed Modified Two-Steps Method (MTSM) from LMI feasibility point of view
Robust fixed-time consensus tracking with application to formation control of unicycles
No full textInternational audienc
Stabilization of switched affine systems with disturbed state-dependent switching laws
No full textInternational audienceIn this paper we investigate the stabilization problem for a class of switched affine systems with a state-dependent switching law. Since the states measurements are in general subject to perturbations and noises, we propose a robust switching law design method. Qualitative conditions for the stability of the closed-loop switched system are given. Stability conditions are also formulated as Linear Matrix Inequalities (LMIs) in order to allow numerical implementations. Results are illustrated by numerical examples in order to show the efficiency of the method and its limits