Towards Abnormal Behavior Detection of Cognitive Impaired People

The assessment of the general health status of people with cognitive impairment, and so the quality of drug therapy, requires reliable information about their daily lifestyle. Especially in the case of Alzheimer's Disease (AD) patients, the patients themselves or their primary caregivers may not be reliable in providing such information. This raises the demand for a new generation of intelligent, situation-aware environments able to collect, process and store information about patients' daily activities. In this paper, a new approach for the detection of abnormal behavior in AD patients is presented. Instead of relying on the identification of deviations from normal behaviors, the approach is based on the specification and runtime verification of the correctness properties of objects, which are visually recognized by means of non-intrusive RGB-D cameras. The Bigraphical reactive system is the formal method adopted to model the correctness properties; whereas, the runtime verification is performed by a specifically designed software component.