Beyond Tabular Data: Enhancing Clinical Trial Prediction with Heterogeneous Graph Neural Networks

Identifier:  TFM:2530

Handle: https://hdl.handle.net/20.500.11797/TFM2530

Authors:  Muñoz Bravo, Alex

Abstract:
This thesis explores enhancing clinical trial prediction by moving beyond traditional tabular data. It proposes a novel approach using Heterogeneous Graph Neural Networks (GNNs) to model the complex and diverse relationships between biomedical entities, such as drugs, indications and trials. By representing these interactions within a rich, interconnected graph structure, this methodology aims to capture intricate patterns that conventional models may miss. The primary objective is to improve the accuracy of predicting clinical trial outcomes, potentially leading to more efficient drug development pipelines and reduced failure rates.