Big Data Analytics for the Daily Living Activities of the People with Dementia

Abstract

Dementia is still an incurable disease that affects a big part of the population nowadays. It affects millions of people worldwide and the number is expected to increase significantly in the next decades. The persons with dementia face difficulties in performing the daily living activities (DLAs) due to movement disorders, poor coordination and memory loss and they need support from family members or health care professionals. In this paper several Big Data techniques are explored for the analysis of the DLAs of the people that have dementia in order to identify behavioral patterns. In particular this paper: (1) presents how the K-Means Clustering algorithm can be used for the identification of the number of types of DLAs performed by a person with dementia in a day, (2) presents how to apply the Collaborative Filtering algorithm for the prediction of the frequency of the DLAs and (3) compares several classification and regression algorithms for the identification of the days with anomalies with respect to a baseline and for the prediction of the durations of the DLAs of the people with dementia using a prototype developed in-house. Two datasets used in the experiments are taken from literature and a third dataset is derived from one of the previous datasets and used as simulated data.