Back to search

Aristotle University Of Thessaloniki (AUTH)

Οι ψηφιακοί βιοδείκτες ως οικολογικά έγκυρα μέσα απομακρυσμένης και μακρόχρονης εκτίμησης της υγείας των ατόμων τρίτης ηλικίας

Abstract

dc:description

Ageing consists one of the biggest societal challenges worldwide. Older adults’ desire of staying healthy, staying at their own home consists an important motive for the development of assistive technologies that will realize it through remote monitoring technologies. Proliferation of low-cost, off-the-shelf IoT devices has led to the implementation of the so called smart homes projects, which are solutions that are mixing the older adults’ physical spaces with monitoring and computing capabilities, allowing the collection of high-frequency daily life data. Methods: This stydy consist of two parts. An experimental one that takes place in a lab space and a real life application of sensor monitoring technology. In particular, an ecologically valid space was created within the Thessaloniki Active & Healthy Ageing Living, where fifteen (15) older adults participated in the pilot testing of the technology. Participants visited the ecologically valid space for almost two weeks, covering eight (8) 1 hour-sessions in total. There, they followed a protocol of typical daily life activities, where several unobtrusive scenario of monitoring were interweaved. No restrictions were imposed with respect to the execution of tasks, thus making sure that the data collection follows an ecologically valid paradigm. Subsequently, senor data collected from 10 participants were statistically correlated to clinical assessment tools. Then, a series of focus groups openly discussing about technology and obtrusiveness were carried out with thirteen participants. Moving to the second part of the study, some adjustments with respect to the technology had to be considered according to the Living Lab study findings. Favorable technology was installed to 5 older womens’ homes for more than a year time resulting to hundreds of thousands data points to be collected. Different modeling techniques were employed, such as statistical modeling, and machine learning, in order to explore clinical added value of digital biomarkers. As an application scenario, the study of modelling emotional disturbances and in particular the one of geriatric depression was carried out. In addition, a longitudinal study about sensor monitoring unobtrusiveness was conducted to examine end-user acceptance. A follow-up interview was carried out with each one of the participants. Finally, we explore the possibility of mapping the results of the data-driven modelling approaches with expert-driven models and knowledge representation schemata, such as the Fuzzy Cognitive Maps (FCMs) in a way that is transparent to the clinicians. Results: With respect to the first part of the study quite a few statistical significant correlations between sensor data and clinical tools were found, e.g. mobility and affective characteristics linked to emotional health and quality of life. With respect to the results of the focus groups these were transcribed and analysed with qualitative methods in order to extract the most significant themes and to categorize them to one of the obtrusiveness framework axes. For the second part o fthe study, a generalized linear mixed prediction model of PHQ-9 was developed utilizing information about TV usage patterns. Random Forests achieved mean accuracy score >80% when given to classify between healthy and depressive cases, while another RF classifier achieved an AUC>90% when having to deal with all sorts of depressive synptoms severity (from mild to severe). The initial FCM model also achieved a high classification rate up to 96% given some synthetic cases provided by experts. Finally, older adults’ longitudinal attitudes revealed a negative stance towards the use of the mirror camera and the smart watch, as well as the Kinect device. The answers to the questionnaire of the lady that left the study at month 12, were compared against the mean value of the rest four particpants to check for any particular reasons of her decision. Significant differences from the mean value of the rest four were found for the Kinect device, the smart watch and the mirror camera. Conclusions: Clinical value of digital biomarkers has been been revealed in many publications, yet their application in longitudinal studies is absent. A great challenge nowadays remains the robust operation of such technologies under real life circumstances, without any restrictions (in the wild) as well as their acceptance from older people, and their doctors, if at any time in the future we wish to integrate such data in the clinical practice.

Degree

thesis:*
Grantor dc:publisher
Aristotle University Of Thessaloniki (AUTH)
Year dc:date
2020

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Billis, Antonis
  • Μπίλλης, Αντώνιος

Subjects

dc:subject × 22

Rights

Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/47534

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Billis, Antonis; Μπίλλης, Αντώνιος. Οι ψηφιακοί βιοδείκτες ως οικολογικά έγκυρα μέσα απομακρυσμένης και μακρόχρονης εκτίμησης της υγείας των ατόμων τρίτης ηλικίας. Aristotle University Of Thessaloniki (AUTH), 2020. http://hdl.handle.net/10442/hedi/47534