No abstract

Purpose Credit scoring datasets are generally unbalanced. The number of repaid loans is higher than that of defaulted ones. Therefore, the classification of these data is biased toward the majority class, which practically means that it tends to attribute a mistaken “good borrower” status even to “very risky borrowers”. In addition to the use of statistics and machine learning classifiers, this paper aims to explore the relevance and performance of sampling models combined with statistical prediction and artificial intelligence techniques to predict and quantify the default probability based on real-world credit data. Design/methodology/approach A real database from a Tunisian commercial bank was used and unbalanced data issues were addressed by the random over-sampling (ROS) and synthetic minority over-sampling technique (SMOTE). Performance was evaluated in terms of the confusion matrix and the receiver operating characteristic curve. Findings The results indicated that the combination of intelligent and statistical techniques and re-sampling approaches are promising for the default rate management and provide accurate credit risk estimates. Originality/value This paper empirically investigates the effectiveness of ROS and SMOTE in combination with logistic regression, artificial neural networks and support vector machines. The authors address the role of sampling strategies in the Tunisian credit market and its impact on credit risk. These sampling strategies may help financial institutions to reduce the erroneous classification costs in comparison with the unbalanced original data and may serve as a means for improving the bank’s performance and competitiveness.

summary Inflammatory optic neuropathy (ON) is a rare event in Behçet's disease (BD). We report herein a series of ten BD Tunisian patients with ON and describe its clinical features among them. A retrospective review of BD patients (International Study Group for BD criteria) was performed. The patients were divided into two groups: those presenting an inflammatory ON, and those none. The diagnosis of inflammatory ON was based on the clinical examination, visual field and visual evoked potentials. We analyzed the characteristics of the two groups. Ten patients (2.3%) presented an inflammatory ON among our 440 patients. Inflammatory ON was inaugural in 8 cases. Clinical manifestations were as follows: blurred vision (7 cases) and periorbital pain (3 cases). In two cases, the patients did not complain from ophthalmological symptoms. The fundus revealed a papilledema (2 cases), papillary pallor (4 cases), and was normal in 5 cases. Visual field realized in only three patients showed a scotoma in all cases. Visual evoked potentials revealed increased latency in all cases. All patients received corticosteroids associated to an immunosuppressive agent. The comparative study between the two groups revealed that inflammatory ON was significantly more associated to neurological involvement (p<0.0001) and that the disease was more severe in the ON group (p<0.0001). Inflammatory ON in BD is rare and may occur at an early stage of the clinical course of the disease. Its prevalence is certainly underestimated. A systematic visual evoked potential may be interesting as a screening tool.

Hypothyroidism is exceptionnally revealed by hypertrophic myopathy, known as Hoffmann syndrome. We report two new cases of severe hypothyroidism presenting as hypertrophic myopathy, occurring in two men of 49 and 57 years-old.

No abstract

Feature selection is an important technique in machine learning and pattern classification. Most existing studies of feature selection are using the batch learning methods. Such methods are not appropriate for real-world applications especially when data arrive sequentially. Recently, this problem is addressed by some feature selection techniques using online learning. Despite the advantages in efficiency of online feature selection methods, they are not always accurate enough whenhandling real world data. In this paper, we address this limitation by the integration of automated negotiation process. We present a novel method based on negotiation theory for online feature selection (ANOFS) and demonstrate its application to several public datasets.

Rationale, aims and objectives :Over the past decades, thrombophilia testing in patients with venous thrombo-embolism (VTE) has increased tremendously. However, the role of inherited thrombophilia (IT) in prediction the risk of recurrence remains controversial. Consequently, it is still unclear whether thrombophilia testing influences decisions regarding duration of anticoagulation in clinical practices. The aim of this study was to evaluate the impact if IT on VTE treatment decisions and on predicting the risk of recurrence. Methods : A retrospective longitudinal study (January 2011-Decembre 2016) including 190 patients with confirmed VTE referred from internal medicine department for inherited thrombophilia screening was carried out. Results : The mean age patients was 40.2 years and the sex ratio (M/F) was 0.78. IT was confirmed in 27.5% of patients. A long term anticoagulation was decided in 51.6% of patients with IT. There was no significant difference in the duration of anticoagulation between patients with or without IT. VTE recurrence was recorded in 26 (13.7 %) patients. The 24 years cumulative incidence of recurrence was 9% in patients with IT and 14% in those without. IT was not associated with increased risk of recurrence after treatment withdrawal (Hazard ratio=1.31 IC(0.47-3.63); p=0.6). Conclusion : In clinical practice, IT did not influence anticoagulation duration and was not associated with a higher VTE risk of recurrence. It seems to be less relevant for decision making than presumed.

No abstract