DiALog project D. Hughes Introduction Longitudinal Discriminant analysis Credible Intervals Statistical Findings Clinical Findings LoDA - A review ◮ To classify a new patient, Ynew = Ynew , 1, . . . , Ynew , R , we can calculate group membership probabilities, Pnew , g = P Unew = g, Ynew , ψ, θ = πg fg . . . G−1 g=0 πg f g . . . ◮ ψg denotes the fixed effects parameters and θg the mixture components in a prognostic group g from the GLMM. ◮ Calculate Pnew , g by averaging across each realisation of the posterior distribution from the MCMC procedure. Pnew , g = 1 M M m=1 Pnew , g ψ(m) , θ(m) , g = 0, . . . , G − 1.