Activity pattern mining using temporal relationships in a smart home

Allowing elders and people with cognitive dysfunction such as Alzheimers disease to stay in their home longer is rapidly becoming a priority for the health care system. The use of smart homes is a very promising solution because it can automatically offer real-time assistance to complete daily activities. Data mining techniques have been used to identify smart home occupant's daily routines, but most of the time, only the sequence of the events is analyzed. We propose a new algorithm to discover frequent activities in a smart home history log using temporal relationships between sensor activations. Experiments on a real smart home sensor log showed promising results in the detection of all frequent activities.