Cancer mortality among French nuclear contract workers.

BACKGROUND: Nuclear workers from French contracting companies have received higher doses than workers from Electricité de France (EDF) or Commissariat à l'Energie Atomique (CEA). METHODS: A cohort study of 9,815 workers in 11 contracting companies, monitored for exposure to ionizing radiation between 1967 and 2000 were followed up for a median duration of 12.5 years. Standardized mortality ratios (SMRs) were computed. RESULTS: Between 1968 and 2002, 250 deaths occurred. Our study demonstrated a clear healthy worker effect (HWE) with mortality attaining half that expected from national mortality statistics (SMR = 0.54, 95% CI = [0.47-0.61]). The HWE was lower for all cancers (SMR = 0.65) than for non-cancer deaths (SMR = 0.46). The analysis by cancer site showed no excess compared with the general population. Significant trends were observed according to the level of exposure to ionizing radiation for deaths from cancer, deaths from digestive cancer and deaths from respiratory cancer. CONCLUSIONS: The mortality of nuclear workers from contracting companies is very low compared to French national mortality.

Multiclass feature selection with kernel Gram-matrix-based criteria.

Feature selection has been an important issue in recent decades to determine the most relevant features according to a given classification problem. Numerous methods have emerged that take into account support vector machines (SVMs) in the selection process. Such approaches are powerful but often complex and costly. In this paper, we propose new feature selection methods based on two criteria designed for the optimization of SVM: kernel target alignment and kernel class separability. We demonstrate how these two measures, when fully expressed, can build efficient and simple methods, easily applicable to multiclass problems and iteratively computable with minimal memory requirements. An extensive experimental study is conducted both on artificial and real-world datasets to compare the proposed methods to state-of-the-art feature selection algorithms. The results demonstrate the relevance of the proposed methods both in terms of performance and computational cost.