Abstract

In this paper, we introduce the use of a personalized Gaussian Process model(pGP) to predict the key metrics of Alzheimer's Disease progression (MMSE,ADAS-Cog13, CDRSB and CS) based on each patient's previous visits. We start bylearning a population-level model using multi-modal data from previously seenpatients using the base Gaussian Process (GP) regression. Then, this model isadapted sequentially over time to a new patient using domain adaptive GPs toform the patient's pGP. We show that this new approach, together with anauto-regressive formulation, leads to significant improvements in forecastingfuture clinical status and cognitive scores for target patients when comparedto modeling the population with traditional GPs.

 

Quick Read (beta)