Understanding home inactivity for human behavior anomaly detection

The importance of Ambient Assisted Living (AAL) systems as a new paradigm to provide assistance to independent older adults at home is constantly growing. In this context, one of the most demanded feature is the possibility to promptly alert caregivers if an anomaly -e.g., a fall or a domestic accident - occurs. However, it is particularly difficult to monitor situations where the subject performs activities that do not involve motion (e.g., "sleeping", "watching TV"). In such cases, there is the need to understand whether the behavior of the person at home has a normal flow or if, for example, it is due to an illness that the monitoring system might not distinguish from other activities. In this work, we deepened the concept of home inactivity and we report a statistical survey study to identify the most relevant variables (observable set) that should be considered while designing an efficient AAL system.