Title: Care Dependency Prediction in Care Homes
Presentation Type: Oral Presentation
Alisa Faingold, Ben Gurion University, Israel
Robert Moskovitch, Ben Gurion University, Israel
Demographic aging, along with a constant increase in life expectancy in modern societies, poses a significant challenge to senior dependency. Dependency is a very complex and multi-domain construct that includes physical, psychological, cognitive, and economic components, making determining the factors that cause dependency to alter more difficult. Being able to predict change in residents’ dependency may enable care homes to better manage their resources, and increase the quality of care. In this study, we predict the level of dependency of care homes’ residents across the UK consisting of a unique database that was collected by Person Centred Software’s Mobile Care Monitoring application, which allows care homes to digitally record various activities in high granularity on a daily basis using data collected from up to 2,800 care homes. We geared novel methods from data science, including specifically temporal abstraction to transform the heterogeneous multivariate temporal data into a uniform representation of symbolic time intervals, and discovered frequent Time Intervals Related Patterns (TIRPs), which are used as features for the prediction models. Our evaluation resulted with prediction accuracy of around 85%, varying for different care homes. Additionally, our investigation shows which patterns are predictive, and which variables or events. We also investigate why specific care homes’ data enables better prediction.
Conference Comments & FeedbackPlace a comment using your LinkedIn profile
Share this Presentation