A multi-scale fuzzy entropy measure for anomaly detection in activities of daily living

Abstract

Anomaly detection in the Activities of Daily Living (ADL) of older adults is essential for healthcare management, to act to avoid prospective problems early and improve this group's quality of life. Once ADLs are recognised, the gathered information will be utilised to detect anomalies in comparison with routine activities. Existing methods provide some limited reliabilities in detecting the anomalous events in ADLs, particularly due to ignoring the changes in individuals' routine. Therefore, it is important to develop an appropriate method or algorithm that can efficiently detect anomalies in older adults' daily activities. The focus of this study is to distinguish and detect anomalies in ADLs based on data gathered from ambient sensors. In this paper, a novel method based on a Multi-scale Fuzzy Entropy measure to discriminate between normal and anomalous cases in ADLs with a high degree of accuracy is investigated. Experimental evaluation is conducted to detect anomalies in ADL data obtained from two different datasets (ADL and CASAS). The experimental results show that the proposed method can detect anomalous instances with a high degree of accuracy. Comparisons with other methods have also offered support to the proposed method.