PhD Position

Data Analysis Engine for Medical Discovery and Prevention

This project will address an area of growing need within medical prevention, especially in the area of cardiology where there is a significant growth in the number of members of the public with cardiac diseases, but their medical condition often remains undiagnosed until it is too late for any preventive steps (e.g., special diet, exercise) to be undertaken.

Using the historical data of patients with cardiac diseases, the main aim of the project is to develop a tool-set, associate methods and models that will allow for the identification of better early indicators of cardiac disease risks.

In terms of the core research, the objective of the project is to develop and validate a Domain Specific Visual Language (DSVL) that can act as a bridge between the mental model of a medical practitioner, the underlying meta-model mined from the data in the domain, and the data processing pipeline that needs to be executed. It is anticipated to have the following goals:

Develop a method (including appropriate tools) that can be used by domain experts (e.g. medical doctors, clinical practitioners) in order to model, manipulate, and analyse raw patient data. This area is likely to involve building an appropriate meta-model to capture the underlying data, and then offer an approach towards performing computations on this data.
Develop a framework that can handle the data processing, and organise into a pipeline that can efficiently perform the required data analysis, and present the results in a format that can be understood and used by medical experts working in the field.
Develop appropriate techniques that can undertake model transformations and intelligently generate a software architecture that can scale to undertake the computations and data analysis (e.g. perform Map-Reduce, or select and apply other types of data analysis, or statistical treatment to the data as appropriate).
Validate the DSVL with medical experts to ensure that the method translates well to their vocabulary, background, and tasks, respectively.