Skip to Main content Skip to Navigation

Tool-based methodologies for developing assisted living services

Abstract : The growing population of older adults gives rise to a need for assistive computing systems that support independent living, to reduce the number of people being transferred to costly care facilities. The goal of assistive computing is to provide context-aware services that assist older adults in all aspects of daily life, for example, monitoring activities such as meal preparation and providing appointment or medication reminders. Despite much progress, the development of assistive services remains a challenge, because of a lack of supporting approaches and tools. This challenge involves: (1) coping with inter-individual variabilities (e.g., home features and user routines and preferences) to deliver tailored services, (2) monitoring activities over long periods of time and (3) enabling care providers and/or professionals in aging to contribute their expert knowledge towardsservice development. This dissertation presents several contributions to this topic. The primary contributions are two iterative methods dedicated to supporting the development of services that monitor activities of daily living (ADLs). Each of these methods is supported by a set of tools for collecting, analyzing and visualizing monitoring data. These tools ensure the agile development of accurate activity recognizers via a stepwise refinement of the analysis of sensor data. The first method, for recognizing ADLs, encompasses the main variations of a target activity by abstracting over descriptions reported by users. Beyond recognizing ADLs, the second method addresses long-term monitoring shortcomings (e.g., sensor failures) and gives health professionals actionable insights into user activities. A final end-user approach is presented, which provides a tool to enable experts in aging to easily define assisted living services in smart homes. The presented methodologies have been applied to an assisted living platform for aging in place, deployed in the home of 140 users. Experimental results show the effectiveness of all the proposed methods. First, the recognition methodology has achieved an accuracy of 80%, rising to 88% when considering the more routinized participants of the experiment. Second, the method for long-term monitoring of ADLs mostly produced the same interpretations as an expert in activity analysis, who manually analyzed the longitudinal sensor datasets. Finally, the findings reveal good usability of the end-user tool, which has been tested by occupational therapists.
Complete list of metadatas

Cited literature [71 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Monday, September 28, 2020 - 3:27:25 PM
Last modification on : Wednesday, October 14, 2020 - 3:46:24 AM


Version validated by the jury (STAR)


  • HAL Id : tel-02950037, version 2



Rafik Belloum. Tool-based methodologies for developing assisted living services. Mobile Computing. Université de Bordeaux, 2020. English. ⟨NNT : 2020BORD0091⟩. ⟨tel-02950037v2⟩



Record views


Files downloads