Trajectories of change in disease outcomes (e.g. PSA in prostate cancer, EDS in Multiple Sclerosis).
Complex inter-relationships which change over time (e.g. fat mass and physical activity, risk factors for CHD)
Methods to examine and correct for bias due to missing data.
Structural equation models, particularly for longitudinal data.
Multilevel models for growth trajectories.
Fractional polynomial and linear spline models.
Time-varying confounding - marginal structural models and g-estimation.
Example PhD topics
Accounting for time-varying confounding in the relationship between exposures and outcomes which change over time (e.g. physical activity and health)
Assessing the sensitivity to unmeasured confounding of cohort analyses
Quantifying associations between outcomes and exposures which vary over the lifecourse (e.g. cumulative effects of poor health on later outcomes)
Comparison of methods (e.g. fractional polynomials, cubic splines, penalised splines, traditional growth models, the SITAR growth model) for approximating patterns of growth/change over time. A variety of different outcomes could be considered here, depending on the applied interests of the student.
Disentangling the effects of growth and absolute weight on later outcomes (e.g. potentially using methods including: partial least squares methods, tracking of exposures over time, latent class methods).