Semi-online neural-Q_leaming for real-time robot learning

Reinforcement learning (RL) is a very suitable technique for robot learning, as it can learn in unknown environments and in real-time computation. The main difficulties in adapting classic RL algorithms to robotic systems are the generalization problem and the correct observation of the Markovian state. This paper attempts to solve the generalization problem by proposing the semi-online neural-Q_learning algorithm (SONQL). The algorithm uses the classic Q_learning technique with two modifications. First, a neural network (NN) approximates the Q_function allowing the use of continuous states and actions. Second, a database of the most representative learning samples accelerates and stabilizes the convergence. The term semi-online is referred to the fact that the algorithm uses the current but also past learning samples. However, the algorithm is able to learn in real-time while the robot is interacting with the environment. The paper shows simulated results with the "mountain-car" benchmark and, also, real results with an underwater robot in a target following behavior

Functionalized nanostructures with application in regenerative medicine

In the last decade, both regenerative medicine and nanotechnology have been broadly developed leading important advances in biomedical research as well as in clinical practice. The manipulation on the molecular level and the use of several functionalized nanoscaled materials has application in various fields of regenerative medicine including tissue engineering, cell therapy, diagnosis and drug and gene delivery. The themes covered in this review include nanoparticle systems for tracking transplanted stem cells, self-assembling peptides, nanoparticles for gene delivery into stem cells and biomimetic scaffolds useful for 2D and 3D tissue cell cultures, transplantation and clinical application.

L'arrêt cardio-respiratoire (ACR) au cabinet médical est-il une réalité et y a-t-il nécessité de s'équiper d'un défribillateur semi-automatique (DSA)?

Introduction: Cette étude a pour but de déterminer la fréquence de survenue de l'arrêt cardio-respiratoire (ACR) au cabinet médical qui constitue un élément de décision quant à la justification de la présence d'un défibrillateur semi-automatique (DSA) au cabinet médical. Matériel et Méthode: Analyse rétrospective des fiches d'intervention pré-hospitalière des ambulances et des SMUR (Service Mobile d'Urgence et de Réanimation) du canton de Vaud (650'000 habitants) entre 2003 et 2006 qui relataient un ACR. Les variables suivantes ont été analysées: chronologie de l'intervention, mesures de réanimation cardio-pulmonaire (RCP) appliquées, diagnostic présumé, suivi à 48 heures. Résultats: 17 ACR (9 _, 8 _) ont eu lieu dans les 1655 cabinets médicaux du canton de Vaud en 4 ans sur un total de 1753 ACR extrahospitaliers, soit 1% de ces derniers. Tous ont motivés une intervention simultanée d'une ambulance et d'un SMUR. L'âge moyen était de 70 ans. Le délai entre l'ACR et l'arrivée sur site d'un DSA était en moyenne de plus de 10 minutes (min-max: 4-25 minutes). Dans 13 cas évaluables, une RCP était en cours à l'arrivée des renforts, mais seulement 7 étaient qualifiées d'efficaces. Le rythme initial était une fibrillation ventriculaire (FV) dans 8 cas et ont tous reçu un choc électrique externe (CEE), dont 1 avant l'arrivée des secours administré dans un cabinet équipé d'un DSA. Le diagnostic était disponible pour 9 cas: 6 cardiopathies, 1 embolie pulmonaire massive, 1 choc anaphylactique et 1 tentamen médicamenteux. Le devenir de ces patients a été marqué par 6 décès sur site, 4 décès à l'admission à l'hôpital et 7 vivants à 48 heures. Les données ne permettent pas d'avoir un suivi ni à la sortie de l'hôpital ni ultérieurement. Conclusions: Bien que la survenue d'un ACR soit très rare au cabinet médical, il mérite une anticipation particulière de la part du médecin. En effet, le délai d'arrivée des services d'urgences nécessite la mise en oeuvre immédiate de mesures par le médecin. En outre, comme professionnel de la santé, il se doit d'intégrer la chaîne de survie en procédant à une alarme précoce du 144 et initier des gestes de premier secours («Basic Life Support»). La présence d'un DSA pourrait être envisagée en fonction notamment de l'éloignement de secours professionnels équipés d'un DSA.

Cooperation in centralised CATI Household Panel Surveys : A Contact-based Multilevel Analysis to Examine Interviewer, Respondent, and Fieldwork Process Effects

In this research, we analyse the contact-specific mean of the final cooperation probability, distinguishing on the one hand between contacts with household reference persons and with other eligible household members, and on the other hand between first and later contacts. Data comes from two Swiss Household Panel surveys. The interviewer-specific variance is higher for first contacts, especially in the case of the reference person. For later contacts with the reference person, the contact-specific variance dominates. This means that interaction effects and situational factors are decisive. The contact number has negative effects on the performance of contacts with the reference person, positive in the case of other persons. Also time elapsed since the previous contact has negative effects in the case of reference persons. The result of the previous contact has strong effects, especially in the case of the reference person. These findings call for a quick completion of the household grid questionnaire, assigning the best interviewers to conducting the first contact. While obtaining refusals has negative effects, obtaining other contact results has only weak effects on the interviewer's subsequent contact outcome. Using the same interviewer for contacts has no positive effects.

Abstracting qualitative information for process supervision

Supervisory systems evolution makes the obtaining of significant information from processes more important in the way that the supervision systems' particular tasks are simplified. So, having signal treatment tools capable of obtaining elaborate information from the process data is important. In this paper, a tool that obtains qualitative data about the trends and oscillation of signals is presented. An application of this tool is presented as well. In this case, the tool, implemented in a computer-aided control systems design (CACSD) environment, is used in order to give to an expert system for fault detection in a laboratory plant

Validation and reproducibility of a semi-quantitative Food Frequency Questionnaire for use in elderly Swiss women.

OBJECTIVE: The principal aim of this study was to develop a Swiss Food Frequency Questionnaire (FFQ) for the elderly population for use in a study to investigate the influence of nutritional factors on bone health. The secondary aim was to assess its validity and both short-term and long-term reproducibility. DESIGN: A 4-day weighed record (4 d WR) was applied to 51 randomly selected women of a mean age of 80.3 years. Subsequently, a detailed FFQ was developed, cross-validated against a further 44 4-d WR, and the short- (1 month, n = 15) and long-term (12 months, n = 14) reproducibility examined. SETTING: French speaking part of Switzerland. SUBJECTS: The subjects were randomly selected women recruited from the Swiss Evaluation of the Methods of Measurement of Osteoporotic Fracture cohort study. RESULTS: Mean energy intakes by 4-d WR and FFQ showed no significant difference [1564.9 kcal (SD 351.1); 1641.3 kcal (SD 523.2) respectively]. Mean crude nutrient intakes were also similar (with nonsignifcant P-values examining the differences in intake) and ranged from 0.13 (potassium) to 0.48 (magnesium). Similar results were found in the reproducibility studies. CONCLUSION: These findings provide evidence that this FFQ adequately estimates nutrient intakes and can be used to rank individuals within distributions of intake in specific populations.

Evaluation of the GlideScope (R) for tracheal intubation in patients with cervical spine immobilization by a semi-rigid collar

Background: In patients with cervical spine injury, a cervical collar may prevent cervical spine movements but renders tracheal intubation with a standard laryngoscope difficult if not impossible. We hypothesized that despite the presence of a semi-rigid cervical collar and with the patient's head taped to the trolley, we would be able to intubate all patients with the GlideScopeR and its dedicated stylet. Methods: 50 adult patients (ASA 1 or 2, BMI ≤35 kg/m2) scheduled for elective surgical procedures requiring tracheal intubation were included. After standardized induction of general anesthesia and neuromuscular blockade, the neck was immobilized with an appropriately sized semi-rigid Philadelphia Patriot® cervical collar, the head was taped to the trolley. Laryngoscopy was attempted using a Macintosh laryngoscope blade 4 and the modified Cormack Lehane grade was noted. Subsequently, laryngoscopy with the GlideScopeR was graded and followed by oro-tracheal intubation. Results: All patients were successfully intubated with the GlideScopeR and its dedicated stylet. The median intubation time was 50 sec [43; 61]. The modified Cormack Lehane grade was 3 or 4 at direct laryngoscopy. It was significantly reduced with the GlideScopeR (p <0.0001), reaching 2a in most of patients. Maximal mouth opening was significantly reduced with the cervical collar applied, 4.5 cm [4.5; 5.0] vs. 2.0 cm [1.8; 2.0] (p <0.0001). Conclusions: The GlideScope® allows oro-tracheal intubation in patients having their cervical spine immobilized by a semi-rigid collar and their head taped to the trolley. It furthermore decreases significantly the modified Cormack Lehane grade.

Uncertainty Quantification Of A Semi-Supervised Support Vector Regression Reservoir Model

Uncertainty quantification of petroleum reservoir models is one of the present challenges, which is usually approached with a wide range of geostatistical tools linked with statistical optimisation or/and inference algorithms. Recent advances in machine learning offer a novel approach to model spatial distribution of petrophysical properties in complex reservoirs alternative to geostatistics. The approach is based of semisupervised learning, which handles both ?labelled? observed data and ?unlabelled? data, which have no measured value but describe prior knowledge and other relevant data in forms of manifolds in the input space where the modelled property is continuous. Proposed semi-supervised Support Vector Regression (SVR) model has demonstrated its capability to represent realistic geological features and describe stochastic variability and non-uniqueness of spatial properties. On the other hand, it is able to capture and preserve key spatial dependencies such as connectivity of high permeability geo-bodies, which is often difficult in contemporary petroleum reservoir studies. Semi-supervised SVR as a data driven algorithm is designed to integrate various kind of conditioning information and learn dependences from it. The semi-supervised SVR model is able to balance signal/noise levels and control the prior belief in available data. In this work, stochastic semi-supervised SVR geomodel is integrated into Bayesian framework to quantify uncertainty of reservoir production with multiple models fitted to past dynamic observations (production history). Multiple history matched models are obtained using stochastic sampling and/or MCMC-based inference algorithms, which evaluate posterior probability distribution. Uncertainty of the model is described by posterior probability of the model parameters that represent key geological properties: spatial correlation size, continuity strength, smoothness/variability of spatial property distribution. The developed approach is illustrated with a fluvial reservoir case. The resulting probabilistic production forecasts are described by uncertainty envelopes. The paper compares the performance of the models with different combinations of unknown parameters and discusses sensitivity issues.

Leishmania amazonensis exhibits phosphatidylserine-dependent procoagulant activity, a process that is counteracted by sandfly saliva

Leishmania parasites expose phosphatidylserine (PS) on their surface, a process that has been associated with regulation of host's immune responses. In this study we demonstrate that PS exposure by metacyclic promastigotes of Leishmania amazonensis favours blood coagulation. L. amazonensis accelerates in vitro coagulation of human plasma. In addition, L. amazonensis supports the assembly of the prothrombinase complex, thus promoting thrombin formation. This process was reversed by annexin V which blocks PS binding sites. During blood meal, Lutzomyia longipalpis sandfly inject saliva in the bite site, which has a series of pharmacologically active compounds that inhibit blood coagulation. Since saliva and parasites are co-injected in the host during natural transmission, we evaluated the anticoagulant properties of sandfly saliva in counteracting the procoagulant activity of L. amazonensis . Lu. longipalpis saliva reverses plasma clotting promoted by promastigotes. It also inhibits thrombin formation by the prothrombinase complex assembled either in phosphatidylcholine (PC)/PS vesicles or in L. amazonensis . Sandfly saliva inhibits factor X activation by the intrinsic tenase complex assembled on PC/PS vesicles and blocks factor Xa catalytic activity. Altogether our results show that metacyclic promastigotes of L. amazonensis are procoagulant due to PS exposure. Notably, this effect is efficiently counteracted by sandfly saliva.

The peri-urban interface and house infestation with Triatoma infestans in the Argentine Chaco: an underreported process?

Peri-urban infestations with triatomine bugs, their sources and their dynamics have rarely been investigated. Here, we corroborated the reported occurrence of Triatoma infestans in a peri-urban area and in neighbouring rural houses in Pampa del Indio, in the Argentine Chaco, and identified its putative sources using spatial analysis and demographic questionnaires. Peri-urban householders reported that 10% of their premises had triatomines, whereas T. infestans was collected by timed manual searches or community-based surveillance in only nine (3%) houses. Trypanosoma cruzi-infected T. infestans and Triatoma sordida were collected indoors only in peri-urban houses and were infected with TcV and TcI, respectively. The triatomines fed on chickens, cats and humans. Peri-urban infestations were most frequent in a squatter settlement and particularly within the recently built mud houses of rural immigrants, with large-sized households, more dogs and cats and more crowding. Several of the observed infestations were most likely associated with passive bug transport from other sources and with active bug dispersal from neighbouring foci. Thus, the households in the squatter settlement were at a greater risk of bug invasion and colonisation. In sum, the incipient process of domestic colonisation and transmission, along with persistent rural-to-urban migratory flows and unplanned urbanisation, indicate the need for active vector surveillance and control actions at the peri-urban interface of the Gran Chaco.