In the context of an intelligent habitat assisting an occupant with Alzheimer?s disease, the goal of plan recognition is to predict the patient?s behavior in order to identify the various ways of supporting him in carrying out his daily activities. However, this situation raises the following dilemma: the observation of a new action, different from the expected one, cannot be directly interpreted as an error; this action can instead constitute the beginning of a second plan, carried out in an interleaved way. In addition, this same action is not inevitably the result of a multiple plan realization; it can effectively be an error. To resolve this dilemma, we propose in this paper a hybrid recognition model based on probabilistic description logic. An implementation of this model was tested in a real smart home infrastructure, by simulating a set of real case scenarios.