Abstract
The aim of this work is to propose new solutions in the field of health event cluster detection. This type of analysis is traditionally used in the survey of diseases of which the aetiology is unknown in order to locate and detect aggregates which have an abnormally high density in time and/or in space. The determination of these clusters generally represents a preliminary stage in search of risk factors.<br />We propose a review of existing methods as well as our contribution in various directions. Two approaches are proposed in the temporal frame. The first allows to avoid the use of simulations. The second makes it possible to take into account data of which the temporal information is incomplete. We have also developped a method for the detection of arbitrarily shaped spatial clusters in order to be able to analyse data of which the exact geographic location is known. This approach has been applied on particular data, those obtained by functional Magnetic Resonance Imaging. The perspectives of spatio-temporal analysis are finally evoked.