Network-based predictive modelling of cardiovascular disease risk

Description

The risk of cardiovascular disease (CVD) is orchestrated by multiple factors. QRISK models (currently QRISK3) have been used in the UK to estimate CVD risk within the next 10 years for individuals without CVD. This helps identify those for whom interventions or more frequent assessment may be needed. QRISK models brought cost benefits for the healthcare system.

QRISK3 model is developed using multivariate Cox proportional hazards modelling (Hippisley-Cox et al. 2017). It explains 59.6% of the variation in time to diagnosis for women, and 54.8% for men.

To improve CVD risk prediction, this project invites you to develop advanced computational modelling utilising a network-based approach. We will model different risk factors as a set of nodes interacting with each other via edges whose activation scores are weighted. A high activation score indicates a strong link between the two nodes, which changes the scores of this pair of nodes. The influence of each node to its neighbours is propagated throughout the network to change the scores of all nodes. We will optimise the activation score of each edge so that the propagated profile can best predict the time to diagnosis. After optimisation, some edges will be removed, while important edges will be highlighted with large activation scores.

The advantage of this network-based approach is it may systematically identify all interactions among factors. We will use a supervised learning algorithm known as network-based supervised stratification (NBS2) to reduce computation time required for model construction (Zhang et al. 2018).

You will gain an impressive set of skills in mathematical modelling for patient selection and efficacy evaluation in drug discovery and development, which is in short supply for the pharmaceutical industry.

We encourage students from quantitative sciences and pharmacology backgrounds to apply. Good B.Sc minimum. Master’s degree desirable.

 

Please note - the listed deadline is provisional. Applications will close once the right candidate is found, so you are encouraged to submit yours as soon as possible.

Availability

Open to students worldwide

Funding information

Self-funded project

The project is open to both European/UK and International students. It is UNFUNDED and applicants are encouraged to contact the Principal Supervisor directly. 

Assistance will be given to those who are applying to international funding schemes. 

The successful applicant will be expected to provide the funding for tuition fees (~£4.6k per year for UK students) and living expenses (~£12k per year). Research costs is minimum. 

A £2000 ISMIB Travel and Training Support Grant may be available to new self-funded applicants who are paying for their own tuition fees

Details of costs can be found on the University website: https://www.liverpool.ac.uk/study/postgraduate-research/fees-and-funding/fees-and-costs/

Supervisors

References

  1. Zhang et al. (2018) Classifying Tumors by Supervised Network Propagation. Bioinformatics. 34(13):i484-i493.
  2. Hippisley-Cox et al. (2017) Development and validation of QRISK3 risk prediction algorithms to estimate future risk of cardiovascular disease: prospective cohort study. BMJ. 357: j2099.