ADL Simulation with Abnormal Behavior of Mild Cognitive Impairment Sufferers

With the increase in the proportion of the elderly population in recent years, age-related health issues have gradually caught more attention. One of them is dementia, which leads to the elderly people having difficulty performing activities of daily life (ADL). There is a phase of mild cognitive impairment (MCI) before the disease progresses to dementia. In the MCI stage, the cognitive deficits are not severe enough to influence ADL. Nevertheless, the disease is progressive and as cognitive deficits accumulate, patients may have increasing problems performing daily activities. Long-term ADL monitoring and abnormal behavior detection can detect early symptoms in incubation periods and assist clinical to evaluate the risk and symptoms of related diseases for early diagnosis, prevention and intervention. However, the collection of real-world ADL sequences is scarce because of the considerable resources and manpower required for long-term monitoring. This study designs an ADL simulator for MCI based on the theory of human behavior and demands. Firstly, the ADL simulator estimates activity attributes and time parameters that is associated with the behavior intention and requirements. After that, the ADL instances are sorted by the order of start time estimated in time parameters. Additionally, the intrinsic structure of activities and the context between sub-activities are important clues for capturing the cognitive status. Therefore, this work also simulates sub-activity-related abnormal behaviors of MCI. The evaluation of ADL simulator shows that the simulated ADL sequences are similar to real-world behaviors and the results demonstrate the feasibility of the proposed method.