What are the Best Methodologies to Support the Analysis of Accelerometry Data for Population Health - Mathematics, PhD (GW4 BioMed MRC DTP)

About
Supervisory Team: Dr Mark Kelson
Location: Exeter, Streatham Campus

Project
Physical activity (PA) is associated with reduced risk of a number of health conditions. Recent advances in wearable technologies have facilitated the collection of large amounts of data recording the PA that people undertake. Current analysis methods for handling this rich data are inefficient and rely on assumptions that may affect the conclusions drawn. This PhD will explore the scale of this problem, the impact of choice of analysis methods on study results (including randomised controlled trials), suggest ways for improvement and provide an accessible platform for implementing those improvements.

There will be four distinct components of the PhD which we expect will form chapters for the eventual thesis.

1.Systematic review of current methods for processing accelerometer data. A formal review of the literature aimed at improving or increasing physical activity measured using an accelerometer will be conducted. This will provide a strong grounding for the student in current accelerometry data analysis practices. This component will be published to provide a summary of available methodologies and their advantages and disadvantages.

2. Development of methods for handling missing data Building on the previous component this chapter will explore the impact of missing data handling in accelerometry analysis (e.g. currently it is routine practise for researchers to set a minimum number of days/hours weartime that constitutes a valid week/day). This section will be motivated by inefficiencies identified in the previous stage but could also include examination of multiple imputation approaches at higher or lower levels of the data structure. Treatment adherence methodologies have been underused in this setting and will be incorporated also.

3. Compositional data analysis Studies aimed at increasing the amount of time spent doing PA face a particular challenge- each day has a fixed and finite amount of time. A study which successfully increases time spent doing PA must by definition also result in a decrease in some other daily activity. This type of data is called compositional data. Treating these measurements as compositional data allows us to identify the impact of interventions not only in terms of what activities were increased, but also at the expense of what other activities. This is a crucial piece of linked information. This section of the PhD will explore the ways in compositional data analysis can enrich the conclusions and interpretations drawn from accelerometry data.

4. Open access A key feature of good science is that it open and reproducible and a key feature of methods research is that it reaches it’s target audience. This final component will consist of packaging and presenting the advancements made in sections 2 and 3 in a usable format. This could include the curation of an open access repository (such as GitHub) or the development of an R package or other approaches that achieve the same goal.

Start date: October 2018

Most studentships will be 3.5 years full time or up to 7 years part-time, and can be longer where additional training is undertaken.


How to apply
APPLICATIONS OPEN ON 25 SEPTEMBER AND CLOSE AT 17:00 ON 24 NOVEMBER 2017.

IMPORTANT: In order to apply for this project, you should apply using the DTP’s online application form. More information on the application process may be found here: http://www.gw4biomed.ac.uk/projects-2/for-students/

You do NOT need to apply to the University of Exeter at this stage – only those applicants who are successful in obtaining an offer of funding from the DTP will be required to submit an application to study at Exeter.