Global demographics show a steady growth in the population of cognitively impaired patients. Consequently, the aging societies are looking to adopt smart technologies in healthcare services to early detect the onset of cognitive decline. These technologies include advanced methods that enable continuous in-house monitoring of the elderly's activities through unobtrusive sensing for recognizing abnormal behaviors that may indicate cognitive deficits. In an earlier work, we proposed a technique to detect the early symptoms of cognitive impairment by continuously monitoring the daily behavior of an elderly at home to recognize fine-grained abnormal behaviors. Recognition was based on rule-based descriptions of anomalies manually defined by domain experts. However, those rules strongly depend on the specific home environment, on the used sensors, and on the particular habits of the elderly; hence, their definition is time-expensive, and rules are not seamlessly portable to different environments. In order to address this issue, in this paper we propose a method to automatically learn the rulebased definitions of behavioral anomalies. In particular, we use a rule induction algorithm to infer those rules based on a dataset of activities and anomalies. We evaluated our method using a dataset of activities and abnormal behaviors carried out in an instrumented smart home. Our method achieves high precision and recall values, around 0.97 and 0.85, respectively, which are comparable to those obtained using manually-defined rules.

Towards automatic induction of abnormal behavioral patterns for recognizing mild cognitive impairment / Z.H. Janjua, D. Riboni, C. Bettini - In: SAC '16 Proceedings of the 31st Annual ACM Symposium on Applied Computing[s.l] : ACM, 2016. - ISBN 9781450337397. - pp. 143-148 (( Intervento presentato al 31. convegno Annual ACM Symposium on Applied Computing tenutosi a Pisa nel 2016 [10.1145/2851613.2851687].

Towards automatic induction of abnormal behavioral patterns for recognizing mild cognitive impairment

Z.H. Janjua
Primo
;
C. Bettini
Ultimo
2016

Abstract

Global demographics show a steady growth in the population of cognitively impaired patients. Consequently, the aging societies are looking to adopt smart technologies in healthcare services to early detect the onset of cognitive decline. These technologies include advanced methods that enable continuous in-house monitoring of the elderly's activities through unobtrusive sensing for recognizing abnormal behaviors that may indicate cognitive deficits. In an earlier work, we proposed a technique to detect the early symptoms of cognitive impairment by continuously monitoring the daily behavior of an elderly at home to recognize fine-grained abnormal behaviors. Recognition was based on rule-based descriptions of anomalies manually defined by domain experts. However, those rules strongly depend on the specific home environment, on the used sensors, and on the particular habits of the elderly; hence, their definition is time-expensive, and rules are not seamlessly portable to different environments. In order to address this issue, in this paper we propose a method to automatically learn the rulebased definitions of behavioral anomalies. In particular, we use a rule induction algorithm to infer those rules based on a dataset of activities and anomalies. We evaluated our method using a dataset of activities and abnormal behaviors carried out in an instrumented smart home. Our method achieves high precision and recall values, around 0.97 and 0.85, respectively, which are comparable to those obtained using manually-defined rules.
Machine learning; Mild cognitive impairment; Recognition of abnormal behaviors; Rule induction; Telecare
Settore INF/01 - Informatica
2016
ACM Special Interest Group on Applied Computing (SIGAPP)
Book Part (author)
File in questo prodotto:
File Dimensione Formato  
16-SAC.pdf

accesso riservato

Tipologia: Post-print, accepted manuscript ecc. (versione accettata dall'editore)
Dimensione 423.27 kB
Formato Adobe PDF
423.27 kB Adobe PDF   Visualizza/Apri   Richiedi una copia
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2434/456526
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 12
  • ???jsp.display-item.citation.isi??? ND
social impact