Use this URL to cite or link to this record in EThOS:
Title: eTRIKS analytical environment : a practical platform for medical big data analysis
Author: Oehmichen, Axel
ISNI:       0000 0004 9356 8681
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
Personalised medicine and translational research have become sciences driven by Big Data. Healthcare and medical research are generating more and more complex data, encompassing clinical investigations, 'omics, imaging, pharmacokinetics, Next Generation Sequencing and beyond. In addition to traditional collection methods, economical and numerous information sensing IoT devices such as mobile devices, smart sensors, cameras or connected medical devices have created a deluge of data that research institutes and hospitals have difficulties to deal with. While the collection of data is greatly accelerating, improving patient care by developing personalised therapies and new drugs depends increasingly on an organization's ability to rapidly and intelligently leverage complex molecular and clinical data from that variety of large-scale heterogeneous data sources. As a result, the analysis of these datasets has become increasingly computationally expensive and has laid bare the limitations of current systems. From the patient perspective, the advent of electronic medical records coupled with so much personal data being collected have raised concerns about privacy. Many countries have introduced laws to protect people's privacy, however, many of these laws have proven to be less effective in practice. Therefore, along with the capacity to process the humongous amount of medical data, the addition of privacy preserving features to protect patients' privacy has become a necessity. In this thesis, our first contribution is the development a new platform called the eTRIKS Analytical Environment (eAE) as an answer to those needs of analysing and exploring massive amounts of medical data in a privacy preserving fashion with the constraint of enabling the broadest audience, ranging from medical doctors to advanced coders, to easily and intuitively exploit this new resource. We will present the use of location data in the context of public health research, the work done in the context of data privacy for location data and the extension of the eAE to support privacy preserving analytics. Our second contribution is the implementation of new workflows for tranSMART that leverage the eAE and the support of novel life science approaches for features extraction using deep learning models in the context of sleep research. Finally, we demonstrate the universality and extensibility of the architecture to other research domains by proposing a model aiming at the identification of relevant features for characterizing political deception on Twitter.
Supervisor: Guo, Yi-Ke Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral