JMIR Publications

ABSTRACT

Background:

Over the past 20 years, various methods have been introduced to construct disease networks. However, established disease networks have not been clinically useful because of differences with demographic factors as well as temporal order and intensity between disease-disease associations.

Objective:

The aim of this study was to investigate the overall patterns of the association between diseases, the network properties such as clustering, degree and strength, and the relationship between the structure of disease network and demographic factors.

Methods:

We used National Health Insurance Service - National Sample Cohort (NHIS-NSC) data from Republic of Korea which include time series insurance information of approximately 2% (1 million) of the total patients between 2002 and 2013. After set observation and outcome period, we choose only 520 common KCD-6 codes that are the most prevalent diagnoses making up approximately 80% of the cases for statistical validity.

Results:

We constructed a directional and weighted temporal network which considered demographic factors and network properties. with RR(Relative Risk) > 4 and FDR-adjusted P-value =.001. Using this data, we were able to obtain disease network with 294 nodes and 3,085 edges. Interestingly, our network shows four large clusters. Analysis of the network topology shows a higher correlation between in-strength and out-strength than in-degree and out-degree. Further, mean age of each disease related to the position along the regression line of the out/in-strength plot. Conversely, clustering analysis revealed that our network has four large clusters with different gender, age, and disease categories.

Conclusions:

We have constructed a directional and weighted disease network visualizing demographic factors. Our proposed disease network model will prove to be valuable tool for early clinical researchers seeking to explore further the relationship between diseases in the future.