Use this URL to cite or link to this record in EThOS:
Title: Behavioural analytics & clinical diagnostics using body sensor networks
Author: Gavriel, Constantinos
ISNI:       0000 0004 7656 9412
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Access from Institution:
All our interaction with the physical world relies on the brain generating movements. However several pathologies, such as Parkinson's disease, adversely affect the way brain controls motor behaviour. Currently, these effects are quantified using various clinical scales, which unfortunately have been proven to be inconsistent as they are primarily based on subjective estimates. Consequently, understanding the effects on human motor control through collection and analysis of high-resolution kinematics is critical for clinical applications and can have a direct impact on neuroscience and biomedical research. We conducted a longitudinal clinical study involving Friedreich's ataxia (FRDA) patients where we deployed a full-body motion capturing suit to collect high resolution kinematic data during standard clinical tests and activities of daily living (ADL). The latter is of fundamental importance as everyday movements in ecologically relevant environments have the potential of telling us a lot more about the onset of a motor disorder than experimentally predefined tasks. The complexity of the collected data required the development of new tools for extracting meaningful information in an objective manner. The statistics of subjects' movement revealed a clear distinction between FRDA patients and control population. These differences compared well with the gold-standard clinical scales, meaning that they can be used as biomarkers to objectively quantify the progression of the disease. Furthermore, we applied unsupervised methods that allowed us to explore the subjects' behavioural differences in a data-driven fashion, even during highly-complex activities. We then explored new ways for monitoring patients within their own personal environment. We designed and implemented the ETHO platform, an ultra-portable and highly affordable set of wearable body sensor network (BSN) nodes which enable high-resolution wireless recordings of real-life human motor behaviour (9DoF inertial sensors) and muscle activations (using our prototype MMG sensors) in unconstrained environments. We verified ETHO effectiveness in monitoring neurodegenerative disorders by collecting the FRDA patients' sleep behaviour on a longitudinal scale and deriving novel biomarkers which objectively highlight the stage of FDRA disease. Our findings verify that the use of BSNs for the continuous monitoring of human behaviour can be a powerful tool for enhancing our understanding of the sensorimotor control. Our statistical tools can be deployed in clinical applications for highlighting the effects caused by various neurological disorders, thus providing more advanced monitoring and objective clinical diagnostic techniques, something that will also allow rapid efficacy measurements during the development of new drugs. Further analysis of kinematic data also has the potential to reveal what methods evolution taught the brain to control a complex and high-dimensional system such as our body. This will drive towards more a new generation of BCI applications, sophisticated robotic controllers and naturalistic prosthetics.
Supervisor: Faisal, Aldo A. Sponsor: Engineering and Physical Sciences Research Council
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral