RFID: Recognizing failures in dressing activity

In elderly care the manual assessment of Activities of Daily Living (ADLs) is a significant problem. Focusing on the dressing we aim not only to recognize the activity performed but also to evaluate its quality, using solely RFID technology. Our approach leverages sparse and noisy RFID readings by applying the two probabilistic methods, Bayesian Networks and Layered HMM. The goal is to follow the garments that patients are manipulating and to detect possible failures such as forgetting to put on something or putting clothes in wrong order. The experimental results showed that both methods can achieve high accuracy.