[en] In this paper we use a well established method for short-term forecasting to predict the amount of hourly Emergency Department (ED) visits in thirteen different hospitals in the south-east area of Tuscany. Our algorithm belongs to the class of similar shape algorithms and perform the forecast in an unsupervised manner. It exploit an historical dataset containing the patient arrival data, in which similar pattern, filtered on the base of a calendar condition, are selected to predict the incoming visit volume for a tunable number of day ahead.