Literatura académica sobre el tema "Uration continue des KGs"

Embedding knowledge graphs (KGs) into continuous vector space is an essential problem in knowledge extraction. Current models continue to improve embedding by focusing on discriminating relation-specific information from entities with increasingly complex feature engineering. We noted that they ignored the inherent relevance between relations and tried to learn unique discriminate parameter set for each relation. Thus, these models potentially suffer from high time complexity and large parameters, preventing them from efficiently applying on real-world KGs. In this paper, we follow the thought of parameter sharing to simultaneously learn more expressive features, reduce parameters and avoid complex feature engineering. Based on gate structure from LSTM, we propose a novel model TransGate and develop shared discriminate mechanism, resulting in almost same space complexity as indiscriminate models. Furthermore, to develop a more effective and scalable model, we reconstruct the gate with weight vectors making our method has comparative time complexity against indiscriminate model. We conduct extensive experiments on link prediction and triplets classification. Experiments show that TransGate not only outperforms state-of-art baselines, but also reduces parameters greatly. For example, TransGate outperforms ConvE and RGCN with 6x and 17x fewer parameters, respectively. These results indicate that parameter sharing is a superior way to further optimize embedding and TransGate finds a better trade-off between complexity and expressivity.

Background: Acute Bilirubin Encephalopathy and kernicterus is an important cause of cerebral palsy, developmental delay and hearing impairment in low-middle income countries. Interventions such as universal screening for neonatal jaundice, Rhesus immunoglobulins, intensive phototherapy and exchange transfusion have made kernicterus rare in high income countries, but in our set up such cases continue to be reported. Methods: Retrospective observational study where case records of term neonates brought to the neonatal ICU with signs and symptoms of acute bilirubin encephalopathy during the years 2016 and 2017 were sought and analysed.Results: A total of ten term babies reported to the neonatal unit with severe hyperbilirubinemia along with signs and symptoms of bilirubin encephalopathy of which 60% were females. 90% had a birth weight of more than 2.5 kg and mean birth weight was 2.7±0.25 kgs. All the babies were out born. A 4 babies were born at home of which 3 pregnancies were completely unsupervised during the antenatal period. 90% of the babies were from the rural areas, 6 of the cases were from the districts Rajouri, Poonch and Reasi where the terrain is hilly, 2 from rural areas of Jammu and 1 from Kathua. Only 1 was from the Jammu city. The age at admission ranged from 3-9 days and serum bilirubin from 24 to 43.3 mg %. A 5 babies had ABO incompatibility, 1 Rh incompatibility, 1 sepsis, while no cause could be found in 3.Conclusions: Neonatal jaundice is often not easily appreciated by mothers and caregivers in the home setting until it becomes severe enough, at which point neurological damage may have already occurred. There is an urgent need to train the primary health care personnel in assessment and early identification of risk factors for severe neonatal hyperbilirubinemia. They can help the families to seek prompt treatment for this preventable cause of cerebral palsy and mental retardation.

Abstract Prophylactic replacement therapy with FVIII is considered the standard of care for patients with hemophilia although a protocol for administering the prophylaxis has yet to be defined by the World Federation of Hemophilia. Gaona in 2012 demonstrated a significant reduction in the number of hospitalizations due to bleeding episodes in patients with Hemophilia A treated with secondary prophylaxis at the Hematology Service of the Hospital Central del Instituto de Prevision Social (HCIPS) with a regimen of 25 IU/Kg per week in two infusions by reporting results from such regimen that had started at the HCIPS in 2010. This is a lower dose to those most frequently used in high resource countries like the Malmo or Utrecht protocols. The goal of this study was to assess the impact of our lower dose prophylaxis regimen on bleeding events, hospitalizations and quality of life as an alternative for low resource countries that cannot afford to provide their patients with higher dose treatments. The Dutch group Haemo-QoL designed the specific questionnaire for adult patients with hemophilia, the Haem-A-QoL with a Spanish version available for patients aged 17 and up. Results: The HCIPS currently has 36 patients receiving prophylactic treatment with FVIII. From these 18 patients consented to filling out the questionnaire. The mean age of the patients was 28. 3 of the patients were married, 2 were in a stable relationship and 13 were single. 62% of patients lived within 20 km of the Hospital and 38% lived further away going up to 200 km. 17 of the 18 patients had the highest level of education possible for their age. Only one patient, the oldest one, had stopped his education at primary school. All of them had severe disease with one or more affected joints. The average FVIII use per year was 68,000 IU ranging from 8,000 IU to 96,000 IU or 917 IU/kg/year. The strongest determinant for receiving less FVIII was missing appointments. Distance to the Hospital was not of significance. Bleeding episodes prior to prophylactic treatment was 2-3 minor traumatic bleeding episodes per week (after brushing their teeth, easy bruising from minor trauma), 2-3 joint bleeds per month at which point they would seek medical assistance and 1-2 major bleeding episodes per year requiring hospitalization. In the last year 5 patients were hospitalized once. One for pneumonia, one for phlebitis, one for dengue fever and two for hemathrosis. They reported a decrease in bleeding episodes at home from 2-3 joint bleeds per month to 1 every other month. Patients reported after traumatic events their bleeding was "normal" (i.e. when brushing their teeth if they started bleeding they would continue bleeding for days at a time whereas when in prophylaxis they would bleed right after the brushing and then it stopped which they took to be normal). Converting the results from the Haem-A-QoL questionnaire to a scoring system of 1-100 being 1 the best QoL and 100 the worst, the average QoL was 52. With scores of 60 for physical health, 50 for feelings, 52 for view of themselves, 48 for work/school, 53 for treatment, 56 for future, 47 for family planning and 45 for dating. Scores for sports and leisure though only amounting to 64, 33% of patients said that category did not apply to them since they did not practice any sports and the dealing category with a score of 37 was the lowest of all. When compared to results published by a Blood Center in Brazil with an on demand regimen patients with severe hemophilia in that study showed a physical health score of 55, 40 for feelings, 35 for self-perception, 60 for sports and leisure, 35 for work and school, 20 for coping, 45 for treatment, 45 for future and 25 for dating with an average overall score of 40. With an annual average usage of 63,683 IU on the severely affected patients. In summary, the proposed dosing of 25IU/kgs/week seems at first glance to reduce bleeding episodes among patients with severe hemophilia as well as hospitalizations. In general patients treated at the HCIPS have a regular to poor quality of life as measured by the Haem-A-QoL questionnaire which contrasts with results from the Brazilian study on quality of life from patients with on demand treatment and roughly same FVIII usage per year showing their patients with severe hemophilia had an overall quality of life from regular to good. Perhaps the psychological factor comes in to play? A follow up study after patients receive proper psychological evaluations may help clarify results. Disclosures No relevant conflicts of interest to declare.

The educational strategies have changed in Pakistan due to COVID-19. This epidemic covers almost 204 countries and territories all over the world and on 11th March 2020, WHO declared this outbreak as a pandemic. Pakistan has become one of the two hundred and four countries over the world has stopped their educational activities and closed their institutions for unspecified d uration. As we know that spring semester was started in the month of January in Pakistani universities, and had covered almost half path of its journey. In this situation, HEC (Higher Education Commission) announced that universities must convert educational activities in hybrid teaching mode and start online teaching. We made a case study on the Learning Management System (LMS), Campus management System, and WhatsApp etc. Apply these tools on private sector engineering university where two hundred and fifty students enrolled in different mathematics courses. The outcome of this study shows that quick switching to the online education system was little bit difficult but successful. Though in the beginning, there were some problems for teachers and students both but overall the experience was good. Government of Pakistan will reopen the education institute in the month of September and in these days we are quite close to resume educational activities as per the regular practice. The aim of this study is to support to continue the hybrid teaching mode at tertiary level in the regular teaching practice specifically of Mathematics courses. In hybrid teaching mode, the ratio of face to face teaching is seventy percent and online tools and teaching strategies used thirty percent. This method of teaching and learning keep students connected with their teachers. Lastly, we suggest that there is still a room for betterment in LMS. The result of this study shows that students’ performance in mathematics class and exam has been improved by using different application, software and gadgets. E-learning teaching and learning method will be beneficial strategy.

Background: Preterm delivery is a major obstetric complication and a leading cause of neonatal mortality and morbidity. In India, the reported incidence of preterm labour is 10-15%. Premature babies are at risk of many immediate and long-term complications. Objectives: With this background, the present study was aimed to analyse the clinical conditions based on maternal, fetal and placental etiology associated with preterm birth. This prospective observational study was conducted in 250 consecutive women who presented with preterm labour between 28 toMethod: <37 weeks in the Department of Obstetrics and Gynaecology at tertiary care hospital during a period from April 2019 to September 2020. Mode of delivery, various maternal risk factors and neonatal outcome were noted. The incidence of preterm births was found to be 16.15%.Results: Maximum number of patients belonged to third gravida group (40.4%). Most common presentation on admission was pain in abdomen (74.8%) and most common associated risk factor was infection (50.8%%) followed by preterm premature rupture of membrane (PPROM) (24.8%). All patients received steroid (betamethasone) before delivery. 90% of patients delivered vaginally and 31.2% of babies required NICU admission. The mean baby weight, mean APGAR Score (5min) and mean NICU stay was 2.41±0.36 kgs, 8.94±0.33 and 9.08±5 days respectively. Conclusion: Preterm labour and birth continue to be a common occurrence, resulting in signicant newborn death and morbidity, as well as nancial burden on families and hospitals. This incidence can be lowered by early identication of proven risk factors with the assistance of universal and appropriate prenatal care. Strengthening the referral system so that high risk patients are managed in tertiary obstetric units with level III NICU facilities will improve the neonatal outcome of these babies.

Les graphes de connaissances centrés sur les entités sont de plus en plus populaires pour recueillir des informations sur les entités. Les schémas des KG sont complexes, avec de nombreux types et prédicats différents pour définir les entités et leurs relations. Ces KG contiennent des connaissances spécifiques à un domaine, mais pour tirer le maximum de ces données, il faut comprendre la structure et les schémas du KG. Leurs données comprennent des entités et leurs types sémantiques pour un domaine spécifique. En outre, les propriétés des entités et les relations entre les entités sont stockées. En raison de l'émergence de nouveaux faits et entités et de l'existence de déclarations invalides, la création et la maintenance des KG est un processus sans fin. Dans cette thèse, nous présentons d'abord une approche destinée à créer un KG dans le domaine de l'audit en faisant correspondre des documents de différents niveaux. Nous introduisons ensuite des méthodes pour la curation continue des KGs. Nous présentons un algorithme pour la fouille des règles conditionnelles et l'appliquons sur de grands KGs. Ensuite, nous décrivons RuleHub, un corpus extensible de règles pour les KGs publiques qui fournit des fonctionnalités pour l'archivage et la récupération des règles. Nous proposons également des méthodes pour l'exploitation des règles logiques dans deux applications différentes: l'apprentissage de règles souples à des modèles de langage pré-entraînés (RuleBert) et la vérification explicable des faits (ExpClaim)
Entity-centric knowledge graphs (KGs) are becoming increasingly popular for gathering information about entities. The schemas of KGs are semantically rich, with many different types and predicates to define the entities and their relationships. These KGs contain knowledge that requires understanding of the KG’s structure and patterns to be exploited. Their rich data structure can express entities with semantic types and relationships, oftentimes domain-specific, that must be made explicit and understood to get the most out of the data. Although different applications can benefit from such rich structure, this comes at a price. A significant challenge with KGs is the quality of their data. Without high-quality data, the applications cannot use the KG. However, as a result of the automatic creation and update of KGs, there are a lot of noisy and inconsistent data in them and, because of the large number of triples in a KG, manual validation is impossible. In this thesis, we present different tools that can be utilized in the process of continuous creation and curation of KGs. We first present an approach designed to create a KG in the accounting field by matching entities. We then introduce methods for the continuous curation of KGs. We present an algorithm for conditional rule mining and apply it on large graphs. Next, we describe RuleHub, an extensible corpus of rules for public KGs which provides functionalities for the archival and the retrieval of rules. We also report methods for using logical rules in two different applications: teaching soft rules to pre-trained language models (RuleBert) and explainable fact checking (ExpClaim)