Search

International audience; This paper focuses on the problem of the hierarchical multi‐label classification of research papers, which is the task of assigning the set of relevant labels for a paper from a hierarchy, using reduced amounts of labelled training data. Specifically, we study leveraging unlabelled data, which are usually plentiful and easy to collect, in addition to the few available labelled ones in a semi‐supervised learning framework for achieving better performance results. Thus, in this paper, we propose a semi‐supervised approach for the hierarchical multi‐label classification task of research papers based on the well‐known Co‐training algorithm, which exploit content and bibliographic coupling information as two distinct papers' views. In our approach, two hierarchical multi‐label classifiers, are learnt on different views of the labelled data, and iteratively select their most confident unlabelled samples, which are further added to the labelled set. The success of our suggested Co‐training‐based approach lies in two main components. The first is the use of two suggested selection criteria (i.e., Maximum Agreement and Labels Cardinality Consistency) that enforce selecting confident unlabelled samples. The second is the appliance of an oversampling method that rebalances the labels distribution of the initial labelled set, which reduces the reinforcement of the label imbalance issue during the Co‐training learning. The proposed approach is evaluated using a collection of scientific papers extracted from the ACM digital library. Performed experiments show the effectiveness of our approach with regards to several baseline methods.

This paper proposes a novel software architecture for developing disease registries. Our proposal consists of five horizontal layers and two transversal layers. It takes into account several aspects of disease registries, such as various sources of data acquisition, security and privacy issues, data analytics, and user management tasks. Several use cases are presented: the Tunisian Fanconi Anemia Registry (TFAR), the Tunisian Gaucher Disease Registry (TGDR), the Tunisian Non-Hodgkin Lymphoma Registry (TNHLR), and the Tunisian Registry of Diabetes Type 1 (TRD-T1).

In this paper, we focus on the hierarchical multi-label classification task of scientific papers, which consists in assigning to a paper the set of relevant classes, which are organized in a hierarchy. The difficulty of manually constructing sufficient labeled datasets renders challenging the automatic classification task of research papers according to hierarchical labels. Multi-view active learning is a widely adopted method to address this issue, by iteratively selecting the most useful unlabeled samples for the multi-view classifiers exploiting disjoint data’ views, and querying an oracle on their real labels. However, none of the state of the art studies in this field is proposed for the hierarchical multi-label classification task. In this paper, we propose an effective multi-view active learning framework for the hierarchical multi-label classification task, applied on scientific papers. Our approach adopts a novel selection strategy that relies on both uncertainty and representativeness criteria when selecting the most informative unlabeled samples in each iteration. Experimental results on a real world dataset of ACM research papers show the efficiency of our approach over several baseline methods.

In response to the researchers need in the bio-medical domain, we opted for automating the bibliographic research stage. In this context, several classification models of supervised machine learning are used. Namely the SVM, Random Forest, Decision Tree, KNN, and Gradient Boosting. In this paper, we conduct a comparative study between experimental results of full article classification and abstract classification approaches. Furthermore, we evaluate our results by using evaluation metrics such as accuracy, precision, recall and F1-score. We observe that the abstract approach outperforms the full article approach in terms of learning time and efficiency.

This article describes how Web services play an important role in several fields such as e-commerce and e-health. As the number of Web services is increasing rapidly, finding the best Web service according to users' requirements becomes more challenging. The traditional method of Web service discovery is based on keyword match. Due to this, many Web services which are most relevant to the user request are left undiscoverable. Some other emergent approaches are based on semantics to improve the quality of the discovered Web services in terms of relevance and satisfaction of user's need. In this paper, the authors present a survey of existing semantic Web services discovery approaches giving priority to relevant ones. Furthermore, this paper provides a critical and comparative analysis of the studied approaches and stands out major challenges to be addressed to substantially enhance the semantic Web service discovery.

Imatinib mesylate (IM) is considered as a highly effective therapy for chronic myeloid leukemia (CML) patients. However, a minority of patients fail to achieve optimal response due to impaired bioavailability of IM. The human organic cation transporter 1 (OCT1; SLC22A1) has been reported to be the main influx transporter involved in IM uptake into CML cells. Genetic variants and/or hOCT1 expression changes may influence IM response. In this study, we aimed to investigate the impact of both hOCT1 polymorphisms located in exon 7 and hOCT1 mRNA levels on the clinical outcome in CML patients. hOCT1 expression profile was determined using the quantitative real-time polymerase chain reaction in 69 CML patients treated with IM (35 responders to IM patients and 34 IM-resistant patients), while genotyping of 69 cases and 51 controls for hOCT1 polymorphisms was performed by direct sequencing after amplification of exon7. Our results showed that the hOCT1 gene was significantly downregulated in the samples of the IM-resistant group when compared with the IM-responder group (p = 0.0211). Moreover, sequencing data show an association in all cases between the SNP 408V>M (g.1222G>A) and an intronic 8 bp (base pairs) insertion of GTAAGTTG (rs36056065) at the 3′ end of exon 7. The genotype and allele distribution of the different SNPs did not differ significantly between the two groups of patients. hOCT1 mRNA expression may serve as a clinical biomarker of response to imatinib and could be useful to predict IM therapy outcome of CML patients.

bcr-abl fusion gene is the hallmark of chronic myeloid leukemia (CML). RQ-PCR provides an accurate measure of the total leukemia cell mass and the degree to which bcr-abl transcripts are reduced by therapy correlates with progression free survival.We report molecular assessment of residual disease in CML Tunisian patients.Between June 2003 and December 2014 we measured bcr-abl mRNA levels in peripheral blood from all Tunisian patients by quantitative real time polymerase chain reaction (RQ-PCR).A total of 708 patients with a mean age of 42 years were included in this study. Based on European Leukemia Net 2013, 80% of the patients achieved an optimal response 20% were in treatment failure. 38% of the patients achieved RM4 which corresponds to a bcr-abl/abl ratio0.01%, 13% of the patients achieved RM4.5corresponding to bcr-abl/abl ratio of 0.0032%.CML patients had a good response to tyrosine kinase inhibitors treatment. RQ-PCR is helpful in detecting any residual disease and determining the depth of the treatment response.

The purpose of disease registries is to collect and analyze data related to specific diseases in terms of incidence and prevalence. Since the data is typically entered by wearable sensors and/or human caregivers, errors in the data fields are often inevitable. In this paper, we propose a new approach to improve data quality in disease registries based on (a) a semi-random combination of parameters and (b) a learning algorithm for detecting and signaling the loss of quality of the entered data. To implement the approach, we have developed a novel adaptive neuro-fuzzy inference system. It is applied to specific sections of the Tunisian Fanconi Anemia Registry with the aims of reducing false alarms and automatically adjusting the parameters of coefficients of the disease. Our experimental results indicate that both aims can be achieved and effectively lead to improved data quality in disease registries.

International audience; Data are limited in developing countries regarding the clinicopathologic features and response to therapy of chronic myeloid leukemia (CML) in the era of imatinib (IM). The objective of this study is to report on the clinicoepidemiologic features of CML in Tunisia, to evaluate the long-term outcome of patients in chronic (CP) or accelerated phase (AP) treated with IM 400 mg daily as frontline therapy, and to determine imatinib's efficacy and safety. From October 2002 to December 2014, 410 CML patients were treated with IM in six Tunisian departments of hematology. Response (hematologic, cytogenetic, and molecular responses) and outcome-overall survival (OS), event-free survival (EFS), and progression-free survival (PFS)-were evaluated. The following prognostic factors were analyzed for their impact on the European leukemia net (ELN) response, OS, EFS, and PFS at 5 years: age, sex, leukocyte count, Sokal score, European Treatment and Outcome Study (EUTOS) score, CML phase, time to starting IM, and impact of adverse events. The median age was 45 years (3-85 years). Two hundred ten (51.2%) patients were male. Splenomegaly was present in 322 of the 410 (79%). Additional cytogenetic abnormalities were encountered in 25 (6.3%) patients. At diagnosis, 379 (92.4%) patients were in CP, 31 (7.6%) were in AP. The Sokal risk was low in 87 (22.5%), intermediate in 138 (35.7%), and high in 164 patients (41.9%). The EUTOS risk was low in 217 (74%), and high in 77 (26%) patients. The rates of cumulative complete cytogenetic response (CCyR), major molecular response (MMR), and molecular response 4/5 log (MR4.5) in CP/AP-CML patients were 72, 68.4, and 46.4%, respectively. The median time to reach CCyR, MMR, and MR4.5 was 6 months (3-51), 18 months (3-72), and 24 months (3-100), respectively. According to the ELN criteria, optimal, suboptimal response, and failure were noted in 206 (51.8%), 61 (15.3%), and 125 (31.4%) patients, respectively. Five-year event-free survival (EFS), progression-free survival (PFS), and overall survival (OS) were 81, 90, and 90%, respectively. By multivariate analysis, AP, high EUTOS risk, and baseline WBC ≥ 150G/l remained independent predictive factors of non-optimal response to IM. The adverse events (AE) of IM were moderate and tolerable. With the caveats that the monitoring of the disease was not optimal, response rates were similar to those reported in previous studies. It is clear to us that improvements should be made in treatment of AP-CML and high Sokal risk group of CP-CML. The frontline use of second-generation tyrosine kinase inhibitor (TKI) is expected to improve the results of the first-line treatment of these high-risk Tunisian patients, but cost and accessibility of this therapy remain the problems in developing countries.

Scientific research teams have immense valuable knowledge that need to be managed. Organizing scientific contributions of team members constitutes a major challenge for the monitoring of knowledge evolution, team member’s competences discovery, and facilitating information retrieval processes. However, performing manual annotations is often time consuming and labor-intensive task, especially in case of complex annotation schemas. Currently, existing knowledge management systems focus on ensuring the scientific knowledge creation, sharing, organization and evaluation but don’t provide a way for helping researchers in the classification task. In this paper, we introduce a knowledge management system that offers an annotation service for researchers’ contributions by including some natural language processing techniques. The provided service process comprises four phases: (1) the semantic enrichment of domain ontology based on the extraction of background data from Babelnet knowledge base, (2) the automatic generation of candidate categories using the enriched domain ontology, (3) the forwarding of the pre-annotated papers to our web-based system to interact with researchers, and finally (4) the human revision of the generated annotations. Evaluation results show its advantage not only in reducing human effort and time consumption during the annotation task but also in improving annotations quality.

Web services have become an industrial standard offering interoperability among various platforms. With their proliferation, it is becoming increasingly difficult to find a web service that satisfies one's requirements. Yet, discovering the most relevant ones is a key solution to benefit from the large number of web services available nowadays on the Internet. This paper presents a practical tool - called rMatcher - to measure the similarity of semantic web services in order to suggest to its user the best ones meeting his requirements. The tool has been developed and experimented on semantic web services described using the OWL-S language.

As it is known to all, there is an imperative need for electronic transfer of healthcare information (patients’ records, laboratory analyses, medical images…) between healthcare organizations to improve the quality level they are offering. However, medical information systems in each organization are different from each other. These different systems compose a heterogeneous, distributed system with high complexity: there are a large number of healthcare related applications that support specific needs effectively but are isolated or incompatible. Thus medical information systems’ interoperability constitutes one of the main challenges of the IT society. Semantic interoperability in healthcare is important especially when all the varying types of data need to interact. Presented in this paper are the most relevant works to our knowledge, dealing with interoperability in medical information systems and a comparative study between the new proposed approaches and research trends to resolve heterogeneity issues.

Acute acalculous cholecystitis is a rare complication in the treatment of acute myeloblastic leukemia. Diagnosis of acute acalculous cholecystitis remains difficult during neutropenic period. We present two acute myeloblastic leukemia patients that developed acute acalculous cholecystitis during chemotherapy-induced neutropenia. They suffered from fever, vomiting and acute pain in the epigastrium. Ultrasound demonstrated an acalculous gallbladder. Surgical management was required in one patient and conservative treatment was attempted in the other patient. None treatment measures were effective and two patients died. Acute acalculous cholecystitis is a serious complication in neutropenic patients. Earlier diagnosis could have expedited the management of these patients., Highlights • Acute acalculous cholecystitis has rarely described in patients undergoing myelosuppressive chemotherapy for acute myeloblastic leukemia. • Acute acalculous cholecystitis associated with poor short term outcome in neutropenic patients. • Acute acalculous cholecystitis should be added to be the causes of acute abdominal pain in neutropenic patients and unexplained fever. • Earlier diagnosis can improve the prognosis of Acute acalculous cholecystitis during chemotherapy for AML.

This paper presents a new approach for fuzzy rule base reduction using similarity concepts and interpolation techniques. The algorithm consists on: First, measure similarity between rules for the best choice of which of them will be deleted. This operation is done without modification of membership functions. Second, if a new input data is presented to the fuzzy system, interpolation techniques will be used to take into account this arriving data. The main idea of this work is to improve accuracy of the fuzzy system after reduction step. A comparative study between three interpolation methods is done. A mathematical case is treated to show the performance of the proposed method.

Whether in vitro antifungal susceptibility fi ndings correlate with the outcome of patients with invasive aspergillosis (IA) remains debated. This study aimed to test whether IA patients ’ outcomes were associated with in vitro susceptibility results. To do so, we determined the in vitro susceptibility to amphotericin B (AMB) of 37 Aspergillus fl isolates from 14 patients with haematological malignancies diagnosed with proven or probable IA, of which 13 were treated with AMB deoxycholate. Minimal inhibitory concentrations (MICs) were determined by Etest with the isolates classifi ed as in vitro sensitive (AMB-S) or resistant (AMB-R) if their MICs were 2 or 2 mg/l, respectively. The association of the patients ’ death with primary disease, administered antifungal treatment, and infection with AMB-R A. fl avus was tested using generalized estimating equations logistic regression. We assessed AMB-R in 31/37 (84%) isolates. In the patients treated with AMB, the survival rate was 2/3 (67%) and 2/9 (22%) for those infected with AMB-S or AMB-R A. fl avus , respectively. Both infection with AMB-R A. fl avus ( P 0.014) strain and acute myelocytic leukaemia as the underlying primary disease ( P 0.036) were independent predictors of death. Our fi ndings indicate that in vitro resistance predicts a poor outcome in patients with A. fl avus invasive disease treated with AMB. Recent advances in non-culture-based microbiological methods should not discourage efforts to obtain in vitro antifungal susceptibility results, which are critical for the choice of antifungal therapy in patients with IA.

Resume Introduction L’anemie de Fanconi (AF) est une maladie hereditaire caracterisee par une heterogeneite genetique et phenotypique. De nombreux groupes ont mis en place des registres propres a leur pays. En Tunisie, le groupe tunisien d’etude de l’AF (GETAF) a cree le registre tunisien de l’AF « TFAR ». Patients et methodes Tous les services d’hematologie et de pediatrie de Tunisie ont ete contactes pour inclure leurs malades diagnostiques entre janvier 1983 et decembre 2008. Le registre TFAR a ete integre dans le site web du groupe GETAF a l’adresse : www.fanconi-tunisie.net . Resultats Le tri des fiches a permis de retenir 142 patients appartenant a 118 familles. L’âge median de diagnostic etait de 11 ans. Une consanguinite etait notee dans 86 % des cas, un syndrome malformatif dans 91 %, une pancytopenie au moment du diagnostic dans 69 %. Parmi 28 patients, 95 % appartenaient au groupe FANCA. Cent neuf patients etaient traites par androgenes et 27 avaient eu une allogreffe de moelle osseuse geno-identique. Un syndrome myelodysplasique (SMD) avait ete diagnostique dans 4 % des cas, une leucemie aigue (LA) dans 6 % et une tumeur solide dans 2 % des cas. L’âge median de survie etait de 17 ans et 5 mois, il etait significativement meilleur chez les patients allogreffes. Conclusion L’AF parait frequente en Tunisie. L’atteinte hematologique est la circonstance de decouverte la plus frequente. Elle est souvent moderee ou severe et necessite un traitement par androgenes ou allogreffe de moelle osseuse. Celle-ci devrait concerner tous les patients ayant un donneur familial HLA-compatible.

Resume Le plasmocytome est une tumeur maligne qui derive des plasmocytes ; sa localisation ovarienne est exceptionnelle ; nous rapportons un cas de plasmocytome ovarien bilateral diagnostique en postoperatoire ; il n’y avait pas de gammapathie monoclonale, ni lyse osseuse et la biopsie osteomedullaire etait normale. La patiente avait recu une radiotherapie postoperatoire ; elle est sans rechute avec un recul de six mois.

In order to select compatible human leucocytes antigens (HLA) donors for bone marrow graft, all the members of 76 families were typed by serology for HLA class I (A and B locus) and class II (DR, DQ locus) by polymerase chain-reaction-sequence-specific primes (PCR-SSP). The HLA typing interpretation revealed the existence of crossing-over in major histocompatibility (CMH) regions for two families, AB and AT, with aplastic bone marrow. The study of crossing-over site has needed the genotyping of seven short tandem repeat (STR) markers located on the short arm of chromosome 6 (D6S291, D6S273, TNFa, C1.2.C, C3.2.11, D6S265, D6S276), using ABI Prism 310 sequencer. HLA and STR Haplotypic analysis enabled us to confirm the crossing-over between locus B and DR in AB family and between locus A and B in AT family. Based in this study, we recommend to be careful in the interpretation of the results of HLA typing between donors and recipients of bone marrow. Complementary investigations should be accomplished for studying genetic abnormalities, which would be involved in this pathology.

The occurrence of an isolated ovarian or pelvic relapse of acute lymphoblastic leukemia (ALL) in complete remission after chemotherapy has rarely been described. We report the case of a 12-year-old girl, treated for ALL, who developed an isolated left ovarian and fallopian tube localization without medullary or blood relapse 4 years after the end of the initial treatment. She presented with an isolated pelvic mass. The diagnosis was established by a CT-guided biopsy. The treatment consisted of a second course of chemotherapy and complementary surgery; a second complete remission was obtained.

Resume Objectif de l’etude Evaluer les resultats therapeutiques des lymphomes gastriques localises agressifs de pronostic favorable apres chimiotherapie et radiotherapie. Patients et methodes Entre fevrier 1993 et decembre 2004, neuf patients atteints de lymphome gastrique primitif localise de haut grade ont ete pris en charge par le comite lymphome du centre hospitalier universitaire de Sfax (Tunisie). Le sex-ratio etait 1,5 et l’âge moyen de 44,9 ans. Le type histologique etait un lymphome B a grandes cellules dans 100 % des cas. La maladie etait selon la classification d’Ann Arbor modifiee de stade IE dans huit cas et de stade IIE dans un cas sans aucun facteur de pronostic defavorable. Le traitement a consiste en une chimiotherapie suivie d’une radiotherapie de l’estomac avec ou sans les aires ganglionnaires perigastrique a la dose de 40 Gy. Resultats A la fin du traitement, tous les patients etaient en situation de remission complete. Avec un recul moyen de 55,7 mois, un patient a ete perdu de vue, les autres etaient en vie en situation de remission complete. Aucune toxicite aigue majeure n’a ete rapportee. Au cours du suivi, aucun patient n’a souffert de toxicite tardive. Le taux de survie globale a cinq ans etait de 100 %. Conclusion L’association de radiotherapie et de chimiotherapie peut etre proposee pour les lymphomes gastriques localises de pronostic favorable, vu les resultats tres satisfaisants en termes de controle de la maladie et de tolerance.

Resume Introduction Les immunosuppresseurs sont habituellement indiques dans le traitement de premiere ligne des formes non severes de l’aplasie medullaire acquise (AMA). Le but de ce travail etait d’evaluer l’efficacite d’un protocole prospectif associant la ciclosporine aux androgenes dans le traitement des AMA non severes de l’adulte. Methodes Vingt-neuf cas diagnostiques et suivis a Sfax ont ete inclus sur une periode de dix ans (1991–2000). Mis a part le traitement symptomatique, tous nos patients ont recu un traitement specifique comportant une bitherapie a base de ciclosporine a la dose de 5 mg/kg par jour et des androgenes a la dose de 0,5 mg/kg par jour. Resultats Les taux de reponses a trois mois, a six mois et a 12 mois sont respectivement de 48, 75 et 87 %. Les taux de survies sont de 52 % a un an et de 37 % a cinq et dix ans. Les principales toxicites etaient hepatiques et renales ainsi que l’hypertension arterielle, qui ont ete observees chez respectivement 53, 16 et 15 % des patients avec une reversibilite respective dans 65, 87 et 100 % des cas. Conclusion Les taux de reponse et de survie dans notre serie sont assez satisfaisants comparativement a ceux obtenus par les immunosuppresseurs (serum antilymphocytaire et ciclosporine) dans la litterature. L’association des androgenes a permis de potentialiser l’immunosuppression induite par la ciclosporine mais au prix d’effets secondaires plus importants.

In this paper we introduce a new approach for adjustment of membership functions, generation, and reduction of fuzzy rule base from data in the same time. The proposed approach consists of five steps: First, generate fuzzy rules from data using Mendel & Wang Method introduced in [1]. Second, calculate the degree of similarity between rules. Third, measure the distance between the numerical values which induces similar rules. Four, if the distance is greater than base value then merge membership functions. Finally, regenerate rules from data with new fuzzy sets. This approach is applied to truck backer-upper control and Liver trauma diagnostic. A comparative study with a simple Mendel Wang method shows the advantages of the developed approach.

We report a case of Lymphocytic Lymphoma presenting with primary manifestation in the prostate. A 82 year-old man presented to emergency department with acute urinary retention. Digital rectal examination revealed a voluminous and firm prostate. Histology confirmed involvement of the prostate by small B Lymphocytic Lymphoma. The patient was treated with chlorambucil. Lymphocytic infiltration of prostate is a rare manifestation. However this may also be the first sign of an undiagnosed lymphoma. This observation shows that the prostatic lymphoma must be considered among the causes of low urinary retention.

The Cloud storage is the most important issue today. This is due to a rapidly changing needs and a huge mass of varied and important data to back up. In this paper, we describe a work in progress and propose a flexible system architecture for data storage in the Cloud. This system is centered on the Data Manager module. This module provides various functions such as the dispersion of data in fragments, encryption and storage of fragments... etc. This architecture proves to be very relevant. It ensures consistency between different components. On the other hand, it ensures the security and availability of data.

Factor XIII deficiency is a rare autosomal recessive congenital disorder of haemostasis characterised by a plasmatic factor XIII level less than 1% in homozygote and bleeding as of the youth. We report a study about ten patients with congenital factor XIII deficiency from seven south-Tunisian families, there are seven females. Umbilical bleeding was common and only two patients had intracranial bleeding. The standard screening tests are normal. Factor XIII activity was less than 1% in all patients. A sub-unit A deficit was detected for the ten patients. Out hemorrhagic context, five patients receive regular prophylactic transfusion with fresh frozen plasma.

Acute myeloid leukemia philadelphie positive (Ph+ AML) is a rare aggressive acute leukemia with poor prognosis. We report a patient with ph positive AML (FAB5), the transcript bcr/abl was not performed at diagnosis. She achieved complete remission after conventional induction chemotherapy. The consolidation therapy was based on Imatinib only due to infectious complications. She was in complete hematologic and cytogenetic remission for 19 months, and after she exhibited an isolated meningeal relapse. A second remission was achieved with intrathecal chemotherapy and cranial irradiation. Imatinib was switched to second generation Tyrosine kinase Inhibitor which had better diffusion into cerebrospinal fluid. She is in complete hematologic, cytogenetic and meningeal remission after 14 months of treatment. Imatinib monotherapy affords insufficient protection from CNS relapse. Second generation Tyrosine kinase Inhibitor seems to have better efficiency. Ph+ AML with monoblastic differentiation should be considered, like Ph+ ALL, at high risk of meningeal leukemia and should receive central nervous system prophylaxis.

Geotrichum capitatum is an uncommon cause of invasive infections in immunocompromised patients, particularly those with hematological malignancies and severe neutropenia. The aim of this study was to report the cases of invasive geotrichosis in our hospital. It is a retrospective study of invasive geotrichosis diagnosed in the Laboratory of Parasitology–Mycology of the UH Habib Bourguiba, Sfax, from January 2005 to August 2013. Six cases of invasive Geotrichum infections were diagnosed. There were three men and three women. The mean age was 35 years. Five patients have acute myeloid leukemia with a profound neutropenia, and one patient was hospitalized in the intensive care unit for polytraumatism. Clinically, the prolonged fever associated with pulmonary symptoms was the predominant symptom (n = 5). Geotrichum capitatum was isolated in one or more blood culture. Two patients had urinary tract infections documented by multiple urine cultures positive for G. capitatum. Five patients received conventional amphotericin B alone or associated with voriconazole. The outcome was fatal in four cases. Invasive geotrichosis is rare, but particularly fatal in immunocompromised patients. Approximately, 186 cases have been reported in the literature. The prognostic is poor with mortality over 50 %. So, early diagnosis and appropriate management are necessary to improve prognosis.