We are looking for a candidate to undertake a project on distributed modelling of cancer outcomes after radiotherapy treatment. This project will use unique and detailed data from hospitals in several countries, and will examine methods for use of routine clinical data to create complex models while maintaining data security across otherwise isolated centres.
Prediction models for cancer outcomes can provide direct clinical value and are becoming increasingly important to individualise treatment. However, some types of cancer are so rare that it can be hard to get sufficient data to create robust models. In addition, ethical considerations and data protection regulations often limit sharing of data between hospitals and thus make data sharing across institutions and countries more challenging. Recently, new data analysis methods, called distributed learning, have opened up for models to be created across several institutions, without data leaving the individual institution.
The project will focus on the radiotherapy-based treatment of anal cancer. However, the methodology around outcome modelling, use of routine clinical data, and distributed learning can also be applied to other cancer sites. Anal cancer can be effectively treated with a combination of radiotherapy and chemotherapy, but some patients still get cancer recurrence, and many experience side effects after treatment. Anal cancer is a rare cancer, which makes single institution datasets too small to be of much use on their own. Additionally, radiotherapy data are complex and multidimensional, making modelling using traditional regression methods challenging. If better models can be created and validated, these can guide optimization of future radiotherapy, making it more effective.
We will connect three centres with anal cancer patient data. As a first step, we will create basic outcome models using distributed learning. Secondly, we will develop methodology to incorporate complex radiotherapy data into the modelling. Alongside these two projects you will work directly with the data managers in the radiotherapy department in Leeds to optimize and streamline local collection of routine clinical data. A final part of the project will depend on your interests, but could involve deep learning using imaging data, creation of decision support tools, or incorporation of patient-reported quality of life outcomes into the framework.
The Leeds Radiotherapy Research Group is a multi-disciplinary group encompassing clinical oncologists, medical physicists, radiographers and statisticians. The group is led by Prof David Sebag-Montefiore and sits within the Leeds Institute of Cancer and Pathology (LICAP), School of Medicine, University of Leeds.
The project will be conducted in close collaboration with Dr Marianne Guren and Prof Eirik Malinen at Oslo University Hospital in Oslo, Norway, and with Prof Andre Dekker and Dr Leonard Wee at the MAASTRO Clinic in Maastricht, Netherlands.
You should hold a first degree equivalent to at least a UK upper second class honours degree in a relevant subject. This project would suit a student with a strong background in either computer science, biostatistics, biomedical engineering, (medical) physics, Machine learning or equivalent. Although not compulsory, the ideal candidate will, in addition to general computer science and biostatistics skills, have prior experience working with health care data.
The project will involve collaborative work, between multiple disciplines, between clinical and academic departments, and across international borders. The candidate will work directly with colleagues in the Leeds Cancer Centre at St James’ University Hospital. They should be willing to undertake one or more stays abroad, including a longer visit to the MAASTRO clinic in The Netherlands, learning the setup and the theoretical background of the distributed learning approach.
The Faculty minimum requirements for candidates whose first language is not English are:
• British Council IELTS - score of 6.5 overall, with no element less than 6.0
• TOEFL iBT - overall score of 92 with the listening and reading element no less than 21, writing element no less than 22 and the speaking element no less than 23.
How to Apply
To apply for this scholarship applicants should complete a Faculty Scholarship Application form using the link below http://medhealth.leeds.ac.uk/download/4087/fmh_scholarship_application_form_2018_19
and send this alongside a full academic CV, degree certificates and transcripts (or marks so far if still studying) to the Faculty Graduate School [email protected]
We also require 2 academic references to support your application. Please ask your referees to send these references on your behalf, directly to [email protected]
by no later than Monday 14 January 2019.
Any queries regarding the application process should be directed to [email protected]