Temporal Graph Convolutional Networks for Predicting Disease Outbreaks in Public Health Surveillance
, Pratheeksha Hegde N, Agitha W, Saravanan M, Subbulakshmi RA new predictive model for disease outbreak prediction is to be developed using Temporal Graph Convolutional Networks (TGCNs).
The emergence and spread of contagious diseases seriously threaten public health systems worldwide. Early diagnosis and disease outbreak prediction are necessary to implement quick solutions and reduce their impacts. Present methods often overlook the complex connections between many factors influencing disease transmission, such as population dynamics, temporal trends, and spatial proximity. To overcome these limitations, we introduce TGCNs as a novel paradigm for disease outbreak prediction.
Temporal convolutional and graph neural networks are coupled in TGCN models to reflect spatiotemporal dependencies in disease spread. The degree of interaction between different locations is shown by edges in a graph, and this spatial relationship forms the foundation of the graph structure. Using this proposed method, temporal convolutional layers, and graph convolutional layers, TGCNs learn temporal patterns in disease incidence data and geographical representations of nodes.
The main research problem in this research is to tackle and develop a prediction model that can accurately predict disease outbreaks based on diverse data sources, which mainly focus on the application of temporal surveillance data and geographical relationship information. In this research work, the objective is to describe the dynamics and underlying structure of disease transmission networks throughout time and to harness the capabilities of graph convolutional networks (GCN), which precisely forecast and focus on future outbreak episodes.
The TGCNs outperformed state-of-the-art methods for disease outbreak prediction using the disease surveillance datasets. Through the effective application of both temporal and spatial information strategies, TGCNs show strong performance across various disease types and geographical areas, which helps to achieve better results and accuracy on an enhanced proposed method.
Investigations should prioritize the validation of the algorithm in various healthcare environments to assess its efficacy in clinical application.
In future research, this work can be enhanced using several deep learning algorithms to achieve better accuracy and performance.
Back