Privacy preserving context-aware framework for cardiac health monitoring

The impact of digital technology on healthcare delivery services is increasing as new technologies evolve and current technologies expand. These technologies have the potential to provide a platform to reason about the health condition of a patient using relevant contextual information. Context-aware reasoning is particularly important in cardiac health monitoring because of the increasing number of deaths resulting from cardiac diseases. As a result, several efforts have been made to develop intelligent systems for cardiac condition monitor-ing. Nevertheless, most of the existing systems for cardiac health monitoring are generally based on physiological information, mainly the heart rate or electrocardiogram(ECG) signals, while the few research that does integrate contextual information has not considered the privacy of the patients in the development process. This research proposes a privacy-preserving context-aware framework for cardiac health monitoring using contextual information from the patient’s behavior data to facilitate physicians’ decision-making. The framework considers patient’s privacy by allowing the user to take control of the data generated from the sensors as information is stored in the user’s device and not transferred to any server. Furthermore, the user’s pri-vacy is also considered at the algorithm training and model generation stage by adopting a federated machine learning technique. Using federated learning for model development which is a key contribution of this research aims to maintain user privacy by allowing clients from different locations to collabo-ratively learn a machine learning model without sending datasets to a central server. In addition, the framework addresses the issue of context acquisition by engaging healthcare professionals in the development process. A prototype tagged ”mCardiac” is presented as a proof of concept. The design, implemen-tation, and evaluation of mCardiac was made possible by constant interaction with healthcare professionals. mCardiac was also evaluated with cardiac pa-tients who were asked to use the system to validate the effectiveness of the approach.

Level dc:type.qualificationlevel

PhD thesis

Grantor dc:publisher.institution

Middlesex University

Year dc:date.issued

2023

Author dc:creator

• Ogbuabor, G.

Identifier

oai:repository.mdx.ac.uk:8q450

OAI identifier oai:identifier

oai:repository.mdx.ac.uk:8q450