SARRIMA: smart ADL recognizer and resident identifier in multi-resident accommodations

Abstract

Systems for measuring Activities of Daily Livings (ADL) play a significant role in home health-care. The ability of performing ADLs successfully is used as an important factor in deciding treatments and services for patients and elderly citizens. However, most of these systems are designed for single-resident homes. The presence of multiple people creates higher numbers of parallel and overlapping activities, and introduces additional complexities in defining and recognizing activity instances. We present SARRIMA, a system that recognizes activity instances and assigns those activities to a person in 2-resident homes using only passive sensors. We evaluate the efficiency of SARRIMA in two different public datasets (data from real homes) with multiple residents. On the average SARRIMA detects more than 97% of the activity instances. We also show how the person assignment accuracy varies as a function of the similarity of behavior of the 2 people living together and of the types of passive sensors installed.