Core Concepts
A novel Functional Graph Convolutional Network (funGCN) framework combines Functional Data Analysis and Graph Convolutional Networks to address multi-task learning complexities in digital health and longitudinal studies.
Abstract
The paper introduces funGCN, a unified approach for handling multivariate longitudinal data in health solutions. It offers task-specific embedding components, classification, regression, forecasting abilities, and a knowledge graph for data interpretation. Simulation experiments validate its efficacy. The EU Horizon 2020 SEURO Project utilizes funGCN for home-based care improvement. Related work lacks a comprehensive framework like funGCN.
Stats
"n = 300"
"p = 20, 50, 100, 500"
"1,518 participants"
"35 variables from EasySHARE dataset"
"General Assembly Resolution et al., 2015"
"80 replications per task"