Authors: Salima Sabri; Abdelouhab Aloui
Addresses: Laboratoire d'Informatique MEDicale (LIMED), Faculté des Sciences Exactes, Université de Bejaia, 06000 Bejaia, Algeria ' Laboratoire d'Informatique MEDicale (LIMED), Faculté des Sciences Exactes, Université de Bejaia, 06000 Bejaia, Algeria
Abstract: The evaluation of a patient's functional ability to perform daily living activities is an essential part of nursing and a powerful predictor of a patient's morbidity, especially for the elderly. In this article, we describe the use of a machine learning approach to address the task of recognising activity in a smart home. We evaluate our approach by comparing it to a Markov statistical approach and using several performance measures over three datasets. We show how our model achieves significantly better recognition performance on certain data sets and with different representations and discretisation methods with an accuracy measurement that exceeds 92% and accuracy of 68%. The experiments also show a significant improvement in the learning time which does not exceed one second in the totality of the experiments reducing the complexity of the approach.
Keywords: ubiquitous applications; automatic learning; Katz ADL; activity recognition; probabilistic models; wireless sensor network.
International Journal of Ad Hoc and Ubiquitous Computing, 2019 Vol.32 No.4, pp.211 - 223
Received: 05 Oct 2017
Accepted: 30 Apr 2018
Published online: 23 Oct 2019 *