PhD opportunity

Dynamic prediction of kidney and cardiovascular complications in type 2 diabetes using joint longitudinal–survival models in linked population health records

Funding availability

Unfunded

Application deadline

31 January 2027

Faculty

Population Health and Genomics , Faculty of Health

  • Funding – China Scholarship Council PhD Programme
  • Application deadline 31st January 2027
  • Standard Entry date - September 2027
  • Applicants are expected to have a degree (equivalent of Honours or Masters) in a relevant discipline.

Type 2 diabetes is a leading cause of chronic kidney disease, kidney failure and cardiovascular disease in Scotland and in China. People with diabetes have kidney function (eGFR), albuminuria, HbA1c, blood pressure and lipids measured repeatedly over many years, yet the risk tools used in clinics take a single snapshot of these values and ignore how they change. A patient whose eGFR is falling steeply carries a different risk from one whose eGFR is low but stable. Prediction that updates each time a new measurement arrives allows kidney- and heart-protective treatments, such as SGLT2 inhibitors and GLP-1 receptor agonists, to be targeted earlier to the people who stand to benefit most.

The research will address four questions:

1. How do the trajectories of eGFR, albuminuria and HbA1c jointly relate to the risk of kidney failure, major cardiovascular events and death?

2. Does a dynamic prediction model, updated at every clinic visit, discriminate and calibrate better than established static risk scores?

3. How should competing outcomes and irregular, health-driven timing of clinic visits be handled when building these predictions?

4. Do the models transport across populations when validated in an independent cohort?

The project uses linked electronic health records for the populations of Tayside and Fife, held in the University of Dundee Health Informatics Centre Trusted Research Environment, bringing together diabetes register data, laboratory results, prescribing, hospital admissions and deaths. The student will build multivariate joint longitudinal–survival models that link several biomarker trajectories to time-to-event outcomes, and extend them to multistate models describing transitions through chronic kidney disease stages to kidney failure, cardiovascular events and death. Estimation will be Bayesian, using scalable approaches designed for large population datasets and building on the joint modelling R packages jmBIG and JMbdirect developed by the principal supervisor. Models will be compared with landmarking and machine-learning survival methods and assessed by time-dependent discrimination, calibration and decision-curve analysis, with external validation in an independent UK primary care cohort.

The student will receive training in longitudinal data analysis, survival and multistate modelling, Bayesian computation, statistical programming in R and reproducible research practice, together with hands-on experience of working with linked health records in a secure environment, including information governance and statistical disclosure control. The student will join an active applied statistics group in Population Health and Genomics within the School of Medicine, present at international conferences, and publish both methodological and clinical papers.

The project connects with diabetes and kidney research in Dundee and with NHS clinical teams in diabetes, nephrology and cardiology, who are the end users of the prediction tools. The methods apply directly to large Chinese diabetes cohorts and electronic health record systems, giving the student a clear route to carry the work forward on return to China.

How to apply

Please apply through the link below, you may contact the principal project supervisor to discuss your interest further, see supervisor details below.

Apply now

Supervisors

Principal supervisor

Second supervisor

Related PhD programme