Please use this identifier to cite or link to this item:
Type: Theses
Title: Preventing Falls in Hospitals with Body Worn Batteryless Sensor Enabled RFID
Author: Shinmoto Torres, Roberto Luis
Issue Date: 2018
School/Discipline: School of Computer Science
Abstract: Falls prevention in older people is a growing area of study. Multiple studies have employed fixed pressure mats on beds and chairs to detect when a person exits the bed or chair and alert caregivers to intervene and supervise the patient. In general, randomized control trials using these techniques found little variation in the number of falls in older people because of the large number of false alarms producing alarm fatigue among caregivers. Wearable technology—capable of providing information from locations other than the bed or chair—provides an alternative to pressure mats used in hospitals. However, studies that use wearable technology for falls prevention or alarming approaches is lacking. Moreover, technological approaches focused on activity recognition rarely consider the activities of older people. This is because collecting data from this population is difficult due to the constraints of their age, especially with hospitalized patients. In addition, previous wearable sensor studies considered battery-powered sensing units to provide consistent measurements. However, the use of batterypowered body-worn sensor units for activity recognition of older people in hospital settings face a number of practical limitations such as the unit’s size, high cost, attachment method and need for maintenance. Furthermore, older people have expressed interest in using wearable sensors built upon RFID technology that are small and unobtrusive. This thesis proposes a falls prevention intervention based on the use of a batteryless (passive) wearable sensor platform capable of identifying patients and their movements; the use of such body-worn, passive sensor has not been previously studied with older people for falls prevention. Despite the benefits of the proposed sensor— small, batteryless, and lightweight—the captured data is usually noisy and sparse, an inherent limitation to human activity recognition problems. Therefore, the main aim of this work is to investigate and evaluate methods for the recognition of activities in older people, in particular, hospitalized older people, wearing a passive wearable sensor as part of a technological falls prevention intervention to generate timely alarms to caregivers to assist patients attempting a high-risk activity. Given the use of a novel sensor, this thesis also investigates the acceptability and wearability of the proposed sensor as perceived by older people. The research in this thesis makes several contributions to the field of human activity recognition. In particular, the formulation and development of methods for human activity recognition to accomplish a technological intervention for the prevention of falls using a novel RFID-based sensor technology, the evaluation of these methods with healthy and hospitalized older populations, the contribution of three datasets for human activity recognition research with different demographics, and the development of a sensor acceptability model to determine acceptability and wearability of the proposed sensor by the trialled cohorts. This thesis suggests that the deployment of a wearable sensor based falls prevention intervention is feasible, especially in a hospital setting. Furthermore, the use of this technology was considered acceptable by the trialed cohorts as it was unobtrusive to physical movements and easy to use; however, older people were conscious of using the device as it was a highly visible sensor prototype.
Advisor: van den Hengel, Anton
Ranasinghe, Damith
Dissertation Note: Thesis (Ph.D.) -- University of Adelaide, School of Computer Science, 2018
Keywords: Older people
falls prevention
wearable sensors
wearable RFID
sensor enabled RFID
machine learning
conditional random fields
ambulatory monitoring
Provenance: This electronic version is made publicly available by the University of Adelaide in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. This thesis may incorporate third party material which has been used by the author pursuant to Fair Dealing exceptions. If you are the owner of any included third party copyright material you wish to be removed from this electronic version, please complete the take down form located at:
Appears in Collections:Research Theses

Files in This Item:
File Description SizeFormat 
Shinmoto Torres2018_PhD.pdfThesis11.66 MBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.