# iGraphCTC: an inter-connected graph convolutional network for comprehensive clinical trial collaborations

## TL;DR

iGraphCTC is a new system using graph networks to help find better collaborators for clinical trials by combining location and treatment data.

## Contribution

The novel integration of multidimensional clinical data into a graph convolutional network for improved collaboration recommendations.

## Key findings

- iGraphCTC improves recommendation accuracy by up to 16.08% in AUC and 14.28% in F1-Score.
- The model outperforms previous approaches by integrating geographical and intervention data.
- Graph-based methods effectively identify potential collaborators in clinical trials.

## Abstract

Pharmaceutical companies are increasingly expanding their global presence by engaging in collaborative clinical research to meet the growing demand for effective chronic disease treatments. However, identifying suitable affiliations and collaboration networks remains a significant challenge. To tackle this, we propose iGraphCTC, a novel framework for clinical trial collaboration that utilizes an adapted Graph Convolutional Network (GCN) to streamline the identification of potential collaborators. The key contribution lies in its ability to integrate multidimensional clinical data (geographical and intervention attributes) into the recommendation process. Based on both geographical and intervention datasets, iGraphCTC achieves maximum improvements of 16.08% (AUC), 14.28% (F1-Score), and 6.68-17.44% (Accuracy@K). These results highlight its capability to enhance recommendation accuracy by addressing limitations of previous models and integrating clinical insights into the recommendation process. Our results demonstrate the effectiveness of graph-oriented approaches in identifying collaborative activities and pinpointing potential collaborators, providing valuable insights into the dynamics of the pharmaceutical industry’s collaborative landscape.