Cookies on this website
We use cookies to ensure that we give you the best experience on our website. If you click 'Continue' we'll assume that you are happy to receive all cookies and you won't see this message again. Click 'Find out more' for information on how to change your cookie settings.
Skip to main content

external supervisor

Thomas Lukasiewicz, Department of Computer Science, University of Oxford

BACKGROUND

Health economic models inform the implementation of new health technologies by evaluating their long-term net effects on health outcomes (e.g. extra survival) and costs. The analytical approaches supporting these analyses typically involve a long-term disease model (e.g. Markov-type or discrete-event simulation models based on statistical models of risks of key disease events/mortality) of a particular structure, combined with evidence about effects of interventions on disease events from clinical trials. The model complexity is increased when aiming to inform decisions for individual patients; the field is looking to increase the flexibility in the analytical approaches.

RESEARCH EXPERIENCE, RESEARCH METHODS AND TRAINING

The project will explore the added value of explainable AI systems to enable more fluid disease risk modelling incorporating a wider range of patient and intervention characteristics as well as increased flexibility in model structure with focus on better informing treatment decisions for individual patients.

A programme of work in assessing cardiovascular disease risk and survival prognosis in middle aged individuals using data from the 500,000 participants recruited in the UK Biobank study and further cohort/routine health data is envisaged. The student will be expected to develop analytical frameworks both using standard statistical/epidemiological methods and, separately, using machine learning/AI methods. A particular aim of the project is to investigate logic-based explanations for the machine learning models such as the importance of input parameters and relevance and the reliability of the rules governing these models.

The performance of the approaches will be compared with respect to individual model components (e.g. reliability in predicting disease events such as heart attacks and strokes) and overall model performance (eg, predicting individual’s life expectancy).

FIELD WORK, SECONDMENTS, INDUSTRY PLACEMENTS AND TRAINING

Training will be provided in statistical and epidemiological methods, machine learning and AI techniques and health economics methods as required. The successful candidate will be expected to have good knowledge/skills in one of the fields of Medical statistics/ Epidemiology/ Disease modelling or Machine Learning/AI methods. Both NDPH and Department of Computer Sciences at the University of Oxford have thriving research communities and the student will benefit from seminars and discussions with other researchers and students. Attending and presenting at scientific meetings will be encouraged.

PROSPECTIVE CANDIDATE

A candidate with an undergraduate degree in Mathematics, Statistics, Computer Sciences or other quantitative discipline and interest in data-driven health research. 

Supervisor