Assessing everyday life behavioral rhythms for the older generation

This paper presents a pattern recognition model for assessing behavioral rhythms in the framework of aging and technologies. The method, previously tried and tested using motion sensors installed in assisted living units, has permitted to establish motion-based behaviors of older-people based on their habits in term of displacements and activity-levels. The method is now expanded to measure more specific patterns of everyday life activity assuming an activity can be pre-identified on the long term using an activity recognition system. The study feasibility, carried out using semi-artificial data, includes an attempt to model disruptive patterns of living linked with dementia. The alert triggering method, part of the model is improved, and has been evaluated using a real-case study to detect behavioral changes with a higher sensibility. A customized software embedding the model shows the potential of the system tool to detect baseline behaviors and changes, which could be related to particular chronic disease symptoms.