Exploring Disease Relationships From Clinical Trial Eligibility Criteria Data Using The Network Approach

Background:Since the dawn of the post genomic era,researchers have adopted a more holistic approach to the study of disease interactions.This era has seen the fusion of biology and network science in order to understand and explore the mechanisms underlying diseases.This study will use network science in order to explore disease relationships from clinical trial data.Objectives:Studying disease interactions is cumbersome because the relationship between diseases is complicated.Diseases can be related to each other through genes,protein interactions,phenotypic symptoms and even metabolic pathways.Since the objective of this study is to further explore disease relationships we hypothesized that we can use clinical trial eligibility criteria data to find the underlying biological mechanisms of disease interactions.We extracted free text metadata from clinical trial data and then constructed a network of diseases using the clinical trial data only.This network representation allows us to visualize diseases while learning their underlying biology.Methods and materials:In order to validate the disease network that was constructed in this research,we compared it to another disease network that was built based on purely genetic data.Results:We found that these two networks had some similarities like both of them having neoplasms as the largest cluster and also having a statistically significant number of identical edges.This similarity shows that it indeed it is possible to find biologically validated disease relationships using clinical trial data.From our disease-disease network,we also found some disease relationships which were not found in the disease gene network but are real life occurrences as evidenced in literature.Our research has provided a way of how to leverage the seemingly superficial clinical trial data in order to learn biological relationships between diseases.