Multi-task longitudinal forecasting with missing values on Alzheimer's Disease

TL;DR

This paper introduces a Bayesian multi-task learning framework using SSHIBA for longitudinal Alzheimer's data, effectively imputing missing values and jointly predicting diagnosis, ventricle volume, and clinical scores, outperforming baselines.

Contribution

It presents a novel application of SSHIBA for joint learning and missing data imputation in longitudinal dementia data, integrating multiple tasks and data views.

Findings

Effective missing value imputation demonstrated.

Outperforms baseline models in multi-task prediction.

Capable of integrating heterogeneous longitudinal data.

Abstract

Machine learning techniques typically applied to dementia forecasting lack in their capabilities to jointly learn several tasks, handle time dependent heterogeneous data and missing values. In this paper, we propose a framework using the recently presented SSHIBA model for jointly learning different tasks on longitudinal data with missing values. The method uses Bayesian variational inference to impute missing values and combine information of several views. This way, we can combine different data-views from different time-points in a common latent space and learn the relations between each time-point while simultaneously modelling and predicting several output variables. We apply this model to predict together diagnosis, ventricle volume, and clinical scores in dementia. The results demonstrate that SSHIBA is capable of learning a good imputation of the missing values and outperforming the baselines while simultaneously predicting three different tasks.