Methods for constructing longitudinal disease progression curves using sparse short- term individual data

Different methodologies for constructing and quantifying the underlying long-term trajectories for Alzheimer's disease (AD) markers are developed and compared using real world and simulated data. A novel four-step approach utilising derivative information on the disease progression, standard non-linear mixed effects models, and a Bayesian approach are examined. The four-step approach and the Bayesian approach provide similar outcomes in terms of disease progression, but the latter deals with sparse data, commonly seen in AD progression, more effectively through the introduction of informative priors. The standard non-linear mixed effects model approach is shown to be inefficient for this purpose.