To prevent the loss of autonomy linked to aging due to physical and / or psychological alterations, new technologies are working to delay its occurrence, detect it, assess it by offering modern and innovative solutions. In this context, our thesis project aims to exploit the contribution of analysis and data processing techniques for monitoring human behavior.This thesis targets two important and complementary parts: the first carries out the daily action recognition performed by a person, to inform us about his degree of autonomy. The second part offers a modern solution to maintain autonomy, it is based on the execution of physical exercices.From a datasets of signals collected by an accelerometer and a gyroscope embedded in a smartphone,...