Structuring coalesced multi-source data for artificial neural network modeling

A cybersociophysical systems project is underway focused on alleviating some of the demands of caregiving of persons with dementia (PWD). In this paper, we explored the possibility of combining behavioral and environmental data to establish an optimum feature set that to assess the pre-agitation period in the PWD> Data are not always collected in the same manner, a caveat that has prevented researchers from exploring the possibility of combining different forms of data. The data used in this study included environmental features such as light, audio, and interior weather conditions. These data were combined with behavioral data from the PWD. In addition, subjective data reported by the caregiver was collected and examined to determine an optimal model to derive user requirements for a notification system that enhances a caregiver's ability to detect indicators of mounting agitation.