Return to search

Supporting lifestories through activity recognition and digital reminiscence

This licentiate thesis discusses how lifelogging technologies can be used to build digital reminiscence systems. Lifelogging is a recent pervasive computing trend where different aspects of someone’s life are captured digitally. The aim of the proposed system is to create digital lifestories that can visualize the life of a person and provide a means for retrieving life experiences. The target users are people with mild dementia who have problems in navigating their daily life and in recalling previous events. The claim is that digital lifestories can be utilized for memory and reminiscence support as well as strengthen the bond between a person with mild dementia and his family. The main focus of the research study is about designing and developing digital reminiscence systems that can be used by people with mild dementia as aiding memory tools. Creating digital lifestories requires capturing of context data, such as places and people, and content data, such as sound and images, using pervasive lifelogging tools. The passive and continues capture of data results in the occurrence of false data and noise. For that, the system should reduce the collected data to not overload the user when reviewing the lifelogs. Another problem is that the life should be segmented in the form of activities that are searchable and accessible. Thus the collected lifelog data should be aggregated and structured into semantic activities and then represented as digital lifestories where context data can be retrieved together with related content. This licentiate thesis proposes solutions for filtering collected data to reduce the user’s efforts when reminiscing. The thesis also presents a method that uses prior knowledge of context data to improve the recognition of activities when creating the digital lifestories. In addition, locations where the user spends significant time can help in determining context parameters such as activities. This licentiate thesis proposes a novel approach that collects and clusters logged locations of the user to improve the activity recognition task. The presented approach defines possible places first, and it then identifies activities based on those places. Images, as content data, are then associated with the activities based on their timeframes so the user can review and adjust the data before saving it to his lifestory. The presented digital reminiscence system was evaluated through a field-test involving 10 people with mild dementia together with their caregivers. Healthcare professionals were also involved in the design and the evaluation of the system to improve the outcome of the study. The preliminary results indicate that the system indeed improves the quality of life for people with mild dementia, as their reminiscence processes are encouraged and that the communication with their surroundings increases in both volume and quality. The thesis shows that digital reminiscence systems, which describe life through activities, can increase the perceived quality of life for people with mild dementia. It also shows that activity recognition can be improved by using prior knowledge of context data and by automatic location clustering.

Identiferoai:union.ndltd.org:UPSALLA1/oai:DiVA.org:ltu-26061
Date January 2011
CreatorsKikhia, Basel
PublisherLuleå tekniska universitet, Datavetenskap, Luleå
Source SetsDiVA Archive at Upsalla University
LanguageEnglish
Detected LanguageEnglish
TypeLicentiate thesis, comprehensive summary, info:eu-repo/semantics/masterThesis, text
Formatapplication/pdf
Rightsinfo:eu-repo/semantics/openAccess
RelationLicentiate thesis / Luleå University of Technology, 1402-1757

Page generated in 0.002 seconds