Graph-Based Stochastic Modeling of the Allee Effect in Tumor Growth Dynamics Using GCN and BERT

Abstract

Tumor growth dynamics exhibit complex behaviors influenced by various 
biological factors, including the Allee effect, which describes a critical population threshold 
below which tumor cells struggle to proliferate. Understanding these stochastic fluctuations 
is crucial for predicting tumor progression and developing effective treatment strategies. In 
this study, we propose a Graph-Based Stochastic Model to analyze the Allee effect in tumor 
growth using Graph Convolutional Networks (GCN) and Bidirectional Encoder 
Representations from Transformers (BERT). The proposed approach constructs a dynamic 
graph representation of tumor cell interactions and employs GCN to capture spatial 
dependencies in tumor progression. Additionally, BERT is leveraged to extract contextual 
information from biomedical literature, enhancing the interpretability of tumor dynamics. 
Our experimental results demonstrate the effectiveness of the proposed model in accurately 
predicting tumor growth patterns while considering stochastic variations introduced by the 
Allee effect. This research offers a novel integration of graph-based deep learning and 
stochastic modeling, paving the way for advanced tumor progression analysis and potential 
applications in personalized oncology.