Mobile application of Artificial Intelligence to vital signs monitoring
: multi parametric, user adaptable model for ubiquitous well-being monitoring

    Student thesis: Doctoral ThesisDoctor of Philosophy

    Abstract

    Over the next decade, current reactive medicine, is expected to be replaced by a
    medicine increasingly focused on wellness so called personalised, predictive, preventive, and participatory (P4) medicine will be less expensive, yet more accurate and effective. It has been reported in the recent literature that wireless medical telemetry has the potential to contribute to this. However, research shows that current level of adoption of wireless medical telemetry in nearly every country is minimal. In addition to the technological challenges, other factors that seem to affect users’ perception and acceptance of the new systems could generally be categorised as: cost efficiency, accuracy and effectiveness of monitoring, as well as security of the services.

    The ultimate goal of this research is to demonstrate that ubiquitous vital signs monitoring of multiple parameters with patient-specific and adaptable inference model can make an accurate individualised prognosis of a patient’s health status and its deterioration. With the ubiquities and individualised approach to the patient as opposed to the occasional, conventional population-based diagnostic flows, we could provide more accurate, cost efficient and effective solution, in order to answer many population-based problems of modern health care systems.

    The framework developed by this research addresses elements that are common to current mobile health monitoring systems such as wireless sensing, data filtering and processing, as well as interconnection of external services, amongst others, but taken into a new integrative level. The model eliminates redundant tasks and therefore reduce cost, time and effort when developing Smart Wearable Systems (SWS) applications and minimising their “time-to-market”. Having a framework such as the one proposed here will allow researchers and
    developers to focus more on the knowledge intrinsic to the patient-relevant data being collected and analysed, as opposed to technical developments and specific programming details.

    The presented model enables adaptive monitoring of patients using patient specific models. This provides a more effective approach to identifying potential health risks and specific clinical symptoms of an individual, particularly when compared to the conventional, population-based diagnostic approaches currently used. It is foreseen that acquisition and analysis of multiple vital sign parameters from a single patient in real time, together with continuous
    adaptation of the level of detail of their analysis will enable a more precise understanding of the patient’s health status and eventual diagnosis goals in the future.

    By designing this highly adaptable, distributed model, capable of self-adjusting over time based on historical vital sign measurements, individuals are able to keep physically active, detection and early notification of potential illness risks is improved, and more accurate treatments “on-the-go” is possible with admission to hospital being reduced. The proposed solution is expected to continue to support illness prevention and early detection, enabling management of wellness rather than illness.
    Date of Award2014
    Original languageEnglish
    Awarding Institution
    • Coventry University
    SupervisorAlexeis Garcia-Perez (Supervisor), Kuo-Ming Chao (Supervisor), Siraj Shaikh (Supervisor) & Raouf Naguib (Supervisor)

    Cite this

    '