Detecting Abnormality in Activities Performed by People with Dementia in a Smart Environment

Abstract – Recognition of abnormalities in the activities performed by users in smart environments has applications in many such fields as surveillance and medical care for differently abled and elderly persons. There are many causes for abnormality in user activities. Missing out some events of an activity, not maintaining the proper order of events and occurrence of irrelevant events are the major causes for abnormality in activities performed by persons with early stages of dementia. In this paper a simple and novel method for identifying missed out events, out of order events and irrelevant events is proposed. The method can be used for measuring how well an activity is performed by the user and for offering corrective suggestions. One of the possible applications of this method is enabling elderly people to lead an independent life. The proposed method was tested using a publicly available test data after introducing abnormalities manually, and is found to produce promising results