Statistical learning of spatiotemporal patterns from longitudinal manifold-valued networks

TL;DR

This paper presents a mixed-effects model for learning and predicting spatiotemporal patterns on networks from longitudinal data, with applications to cortical atrophy in Alzheimer's Disease.

Contribution

It introduces a novel mixed-effects model combined with MCMC-SAEM for analyzing longitudinal manifold-valued network data, capturing individual variability.

Findings

Model accurately describes cortical atrophy propagation.

Personalized predictions of cortical thickness maps.

Variability in atrophy trajectories related to age and disease progression.

Abstract

We introduce a mixed-effects model to learn spatiotempo-ral patterns on a network by considering longitudinal measures distributed on a fixed graph. The data come from repeated observations of subjects at different time points which take the form of measurement maps distributed on a graph such as an image or a mesh. The model learns a typical group-average trajectory characterizing the propagation of measurement changes across the graph nodes. The subject-specific trajectories are defined via spatial and temporal transformations of the group-average scenario, thus estimating the variability of spatiotemporal patterns within the group. To estimate population and individual model parameters, we adapted a stochastic version of the Expectation-Maximization algorithm, the MCMC-SAEM. The model is used to describe the propagation of cortical atrophy during the course of Alzheimer's Disease. Model parameters show the variability of this average pattern of atrophy in terms of trajectories across brain regions, age at disease onset and pace of propagation. We show that the personalization of this model yields accurate prediction of maps of cortical thickness in patients.