Clinical data warehouse watermarking: impact on syndromic measure Guillaume BOUZILLEabcd,1 , Wei PANe, Javier FRANCO-CONTRERASef, Marc CUGGIAabcd, and Gouenou COATRIEUX e a INSERM, U1099, Rennes, F-35000, France b Université de Rennes 1, LTSI, Rennes, F-35000, France c CHU Rennes, CIC Inserm 1414, Rennes, F-35000, France d CHU Rennes, Centre de Données Cliniques, Rennes, F-35000, France e Institut Mines-Telecom;Telecom Bretagne; Latim Inserm UMR1101, Brest, France. f WaToo, Brest, France Abstract. Watermarking appears as a promising tool for the traceability of shared medical databases as it allows hiding the traceability information into the database itself. However, it is necessary to ensure that the distortion resulting from this process does not hinder subsequent data analysis. In this paper, we present the preliminary results of a study on the impact of watermarking in the estimation of flu activities. These results show that flu epidemics periods can be estimated without significant perturbation even when considering a moderate watermark distortion. Keywords. Traceability, Clinical data warehouse, Watermarking

1

Guillaume Bouzillé, LTSI, équipe-projet Données Massives en Santé - Université de Rennes 1, Campus de Villejean - Bâtiment 6, 35043 Rennes Cedex - France, ; E-mail: [email protected].