Behavioral and Environmental Sensing and Intervention for Dementia Caregiver Empowerment
Description
Summary: This dataset includes: (1) time-series data of the ambient environment in houses with person-with-dementia (PWD), (2) three-axis accelerometer time-series data of PWD living at home, and (3) dementia caregiver assessment of PWD behavior/agitation information. Description: Behavioral and Environmental Sensing and Intervention study (BESI) is an experimental, prospective study on community-dwelling person-with-dementia (PWD) and caregiver dyads residing in their own homes. PWDs suffer from memory loss, cognitive, visual, and vocal impairments. These symptoms progressively worsen over time and patients often express agitated behaviors. Agitation in dementia is described as a set of either repetitious behaviors like pacing, verbal repetition, etc., or socially inappropriate and aggressive behaviors such as verbal outbursts or hitting. Caring for a PWD can be physically and emotionally taxing for caregivers. The objective of this study is to detect and predict agitation at an early stage and to notify the caregiver with personalized interventions to prevent escalation. The study aims to empower caregivers, reduce burden, and increase self-efficacy by providing just-in-time interventions. The BESI system is deployed in houses with PWD and CG dyads and generates behavioral and environmental data using the system components. The BESI system consists of smart-watches (Pebbles) as wearables (collecting PWD behavioral data), in-home environmental sensors (collecting room-level environmental data), and user interfacing devices collecting CG’s inputs. These parameters can be combined to extract contextual agitation stimuli such as changes in the environment. Pebble continuously senses the wrist and arm movement patterns of the patient. These data contain 3-axis (x,y, and z-axis) components of acceleration of the wearer’s motion that can be used to extract important features related to physical agitation behavior. The room-level environmental sensors collect light-level (lux), room temperature (°C), relative humidity (%), air-pressure (Pa), and audio features. These environmental data are stored as time-series data. Lastly, the user interfacing devices (a tablet, and a Pebble smart-watch) collect CG inputs. The CG inputs are agitation timestamps, agitation event information, recent PWD activity, PWD mood, and CG mood. These CG inputs are stored as comma-separated-values (CSV) files. The detailed description of this dataset can be found in the readme file below.