Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Exploiting Individual Graph Structures to Enhance Ecological Momentary Assessment (EMA) Forecasting

Mandani Ntekouli, Gerasimos Spanakis, Lourens Waldorp, Anne Roefs

TL;DR

GNNs, by incorporating additional information from graphs reflecting the inner relationships between the variables, notably enhance the results by decreasing the Mean Squared Error (MSE) to $0 .84$ compared to the baseline LSTM model at $1 .02.$

Abstract

In the evolving field of psychopathology, the accurate assessment and forecasting of data derived from Ecological Momentary Assessment (EMA) is crucial. EMA offers contextually-rich psychopathological measurements over time, that practically lead to Multivariate Time Series (MTS) data. Thus, many challenges arise in analysis from the temporal complexities inherent in emotional, behavioral, and contextual EMA data as well as their inter-dependencies. To address both of these aspects, this research investigates the performance of Recurrent and Temporal Graph Neural Networks (GNNs). Overall, GNNs, by incorporating additional information from graphs reflecting the inner relationships between the variables, notably enhance the results by decreasing the Mean Squared Error (MSE) to 0.84 compared to the baseline LSTM model at 1.02. Therefore, the effect of constructing graphs with different characteristics on GNN performance is also explored. Additionally, GNN-learned graphs, which are dynamically refined during the training process, were evaluated. Using such graphs showed a similarly good performance. Thus, graph learning proved also promising for other GNN methods, potentially refining the pre-defined graphs.

Exploiting Individual Graph Structures to Enhance Ecological Momentary Assessment (EMA) Forecasting

TL;DR

GNNs, by incorporating additional information from graphs reflecting the inner relationships between the variables, notably enhance the results by decreasing the Mean Squared Error (MSE) to compared to the baseline LSTM model at

Abstract

In the evolving field of psychopathology, the accurate assessment and forecasting of data derived from Ecological Momentary Assessment (EMA) is crucial. EMA offers contextually-rich psychopathological measurements over time, that practically lead to Multivariate Time Series (MTS) data. Thus, many challenges arise in analysis from the temporal complexities inherent in emotional, behavioral, and contextual EMA data as well as their inter-dependencies. To address both of these aspects, this research investigates the performance of Recurrent and Temporal Graph Neural Networks (GNNs). Overall, GNNs, by incorporating additional information from graphs reflecting the inner relationships between the variables, notably enhance the results by decreasing the Mean Squared Error (MSE) to 0.84 compared to the baseline LSTM model at 1.02. Therefore, the effect of constructing graphs with different characteristics on GNN performance is also explored. Additionally, GNN-learned graphs, which are dynamically refined during the training process, were evaluated. Using such graphs showed a similarly good performance. Thus, graph learning proved also promising for other GNN methods, potentially refining the pre-defined graphs.
Paper Structure (25 sections, 1 equation, 3 figures, 3 tables)

This paper contains 25 sections, 1 equation, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Forecasting on EMA Data
  4. Application of GNN on MTS
  5. Methodology
  6. EMA Data
  7. Forecasting Problem
  8. GNN Models
  9. Graph Construction
  10. EMA Dataset
  11. Experimental Setup
  12. Experiment A: Investigating Different GNNs
  13. Experiment B: Investigating the Efficacy of Different Graph Structures
  14. Experiment C: Investigating Static and GNN-Learned Graphs
  15. Model Parameters
  16. ...and 10 more sections

Figures (3)

  • Figure 1: Personalized forecasting models, repeated across $N$ individuals, aiming at accurately predicting the 1-lag future responses of all $V$ variables.
  • Figure 2: Proposed experimental framework for investigating the effect of different static graphs and graph learning on the forecasting performance of various GNN models.
  • Figure 3: MSE distributions across all individuals comparing the graph learning process to the four static graphs. Apart from the boxplot properties, mean values are given in black as well as the relative percentage of change in red.