Up-to-date prognostic modelling of more detailed population-based cancer data

This is an exciting era for cancer epidemiology in the UK with cancer data quantity and quality increasing. The use of this data should improve the understanding of prognosis following a cancer diagnosis with information on diagnostic procedures, tumour characteristics, treatment and clinical outcomes. However, there are many challenges for the appropriate development and application of statistical methods, which will be the focus of this PhD project. We will utilise a detailed colorectal cancer dataset to develop and implement our new methodology. Three specific areas of interest are outlined below:

1) Multiple imputation with time-dependent effects. As data resources become more detailed, there is also the prospect of an increasing amount of missing data across variables of interest. This part of the project will involve evaluating and developing a number of techniques for building appropriate imputation models.
2) Model selection in large data sets. With increasingly large datasets available just about everything becomes statistically significant without necessarily being clinically important. With a prognostic model the key question is whether making the model more complex alters the relevant predicted probabilities of survival. This part of the project will explore and develop a range of methods to select and compare these complex models.
3) Providing up-to-date survival estimates. Survival estimates for longer-term follow-up are by nature out-of-date as they relate to patients diagnosed long enough ago to have the sufficient follow-up. This part of the project will involve researching and developing approaches to provide up-to-date estimates from a complex prognostic model.

The student would be expected to attend a specialist course on relevant methods (e.g. www.cansurv.net) and present their work at relevant biostatistical and cancer epidemiology conferences. Research visits to the University of Leeds and the Karolinska Institutet in Stockholm, Sweden, will also be possible for the successful applicant.

Applications are accepted until 12 noon on 7th February 2017

In order to apply please select Health Sciences Research for a September 2017 intake on the application form linked below. In the Funding section of the online application, please select Studentship and then Select CMBSP in the next dropdown option. Please indicate the name of the project supervisor(s) and the project title in the space provided within the application.