About the Project
Clinical prediction models (CPMs) take what we know about a person and predict the probability of subsequent outcomes using a regression model or algorithm [1]. An example is QRISK, which uses a Cox model to predict risk of future cardiovascular events given a patient’s weight, blood pressure, cholesterol levels etc [2]. When developing prediction models, little attention is typically paid to causality. In other words, the coefficients in the underlying regression models do not have a causal interpretation. This is undesirable for a number of reasons. First, CPMs based on underlying causal interpretations are known to generalise better to new populations (they have better external validity) [3]. Second, it does not allow us to (correctly) ask ‘what if’ questions, for example, comparing the cardiovascular risk of a patient with and without smoking. As one paper puts it: ‘predictive algorithms inform us that decisions have to be made, but they cannot help us make the decisions.’ [4].
This PhD will develop methods to incorporate causal inference in prediction models, and study the opportunities and challenges that this brings. Specifically, the successful student will build on existing work by our group where we showed the value of using causal inference to ensure that CPMs provide ‘treatment naïve’ risks [5]. This idea can naturally be extended to build models that allow the calculation of risk under a range of possible treatment options, which will allow a more accurate discussion of risk with the patient and clinician, and support informed decisions about treatment planning. The student will also explore the extent to which enriching a CPM with causal principles improves generalisability across populations, and potentially make accurate predictions. Finally, the project will explore how to resolve potential ‘conflict’ between inclusion of a predictor that is highly predictive, but introduces bias into the estimation of causal effects of interest.
Applications are invited from UK/EU nationals only. Applicants must have obtained, or be about to obtain, at least an upper second class honours degree (or equivalent) in a relevant subject.
On the online application form select PhD Cell Biology. For information on how to apply for this project, please visit the Faculty of Biology, Medicine and Health Doctoral Academy website (https://www.bmh.manchester.ac.uk/study/research/apply/)
Funding Notes
EPSRC DTP studentship with funding for a duration of 3.5 years to commence in September 2020. The studentship covers UK/EU tuition fees and an annual minimum stipend £15,285 per annum. Due to funding restrictions, the studentship is open to UK and EU nationals with 3 years residency in the UK.
As an equal opportunities institution we welcome applicants from all sections of the community regardless of gender, ethnicity, disability, sexual orientation and transgender status. All appointments are made on merit.
References
E. W. Steyerberg, “Clinical prediction models: a practical approach to development, validation, and updating.” New York, NY, Springer, 2009.
Hippisley-Cox, J., Coupland, C., & Brindle, P. (2017). Development and validation of QRISK3 risk prediction algorithms to estimate future risk of cardiovascular disease: prospective cohort study. BMJ, 357.
Steyerberg, E. W., Moons, K. G., van der Windt, D. A., Hayden, J. A., Perel, P., Schroter, S., … others. (2013). Prognosis research strategy (PROGRESS) 3: prognostic model research. PLoS Medicine, 10(2), e1001381.
Hernán, M. A., Hsu, J., & Healy, B. (2018). Data science is science’s second chance to get causal inference right: A classification of data science tasks. CHANCE, 32(1), 42–49.
Sperrin, M., Martin, G. P., Pate, A., Van Staa, T., Peek, N., & Buchan, I. (2018). Using marginal structural models to adjust for treatment drop-in when developing clinical prediction models. Statistics in Medicine, 37(28), 4142–4154..
Continue with Facebook
