Por favor, use este identificador para citar o enlazar este ítem: https://doi.org/10.1016/j.knosys.2017.07.025.

Título: Monitoring elderly people at home with temporal Case-Based Reasoning
Fecha de publicación: jul-2017
Cita bibliográfica: Knowledge-Based Systems,Volume 134, 2017, Pages 116-134
ISSN: 0950-7051
Materias relacionadas: CDU::0 - Generalidades.::00 - Ciencia y conocimiento. Investigación. Cultura. Humanidades.::004 - Ciencia y tecnología de los ordenadores. Informática.
Palabras clave: Case-based reasoning; Case-base maintenance; Smart homes
Resumen: This paper presents a study of why and how Case-Based Reasoning (CBR) can be used in the long term to help elderly people living alone in a Smart Home. The work focuses on the need to manage the temporal dimension and how the system must be maintained. The proposal involves the integration of a CBR system in a commercial Smart Home architecture that includes sensors, data communication and data integration. The CBR system analyses the daily activity at home as temporal event sequences, using temporal edit distance to identify the most similar cases. Most common Case-Based Maintenance (CBM) algorithms adapted to the temporal problem (t-CNN, t-RENN, t-ICF, t-DROP1 and t-RCFP) are used to reduce the number of cases in the case base in order to contribute to its long term maintenance. The experiments carried out analyse the effect of different temporal CBM algorithms in common risk scenarios (waking up during the night, falls and falls with loss of consciousness). Data experiments are generated synthetically based on real behaviour patterns of 12 hours’ and 24 hours’ duration. Algorithms are compared using a paired t-test analysis. The results show that the algorithms t-CNN and t-DROP1 are able to create case-bases that statistically present the same average results as the original case-base but with a 10–20% in size. Algorithms t-ICF, t-RCFP and t-RENN can build similar case-bases to the original with a 10–50% size reduction, although they are not totally equivalent since they have significantly different average results than the original case-base. Finally, algorithm t-RENN does not significantly reduce the size of the case-base because it commonly deletes cases describing abnormal scenarios. We demonstrate that the proposed temporal CBR system is able to detect the different proposed risk scenarios when there is a large number of cases. That is, the CBR systems are useful in the long term. Experiments indicate that the temporal CBM algorithms analysed are able to reduce case-bases successfully to detect abnormal scenarios. However, success in creating a maintained case-base equivalent to the original depends on the number of cases.
Autor/es principal/es: Lupani, Eduardo
Juarez, Jose M.
Palma, Jose
Marin, Roque
Facultad/Departamentos/Servicios: Facultades, Departamentos, Servicios y Escuelas::Facultades de la UMU::Facultad de Informática
URI: http://hdl.handle.net/10201/107021
DOI: https://doi.org/10.1016/j.knosys.2017.07.025.
Tipo de documento: info:eu-repo/semantics/article
Número páginas / Extensión: 37
Derechos: info:eu-repo/semantics/openAccess
Attribution-NonCommercial-NoDerivatives 4.0 Internacional
Descripción: © <2017>. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ This document is the Accepted Manuscript version of a Published Work that appeared in final form in [Knowledge-Based Systems].
Aparece en las colecciones:Artículos: Ingeniería de la Información y las Comunicaciones

Ficheros en este ítem:
Fichero Descripción TamañoFormato 
Green-Digitum-MonitoringTCBR-LupianiEtAl_3submission.pdf1,58 MBAdobe PDFVista previa

Este ítem está sujeto a una licencia Creative Commons Licencia Creative Commons Creative Commons