An Epidemiological Human Disease Network Derived from Disease Co-occurrence in Taiwan

Abstract

In "classic" biomedical research, diseases have usually been studied individually. The pioneering human disease network (HDN) studies jointly consider a large number of diseases, analyse their interconnections, and provide a more comprehensive description of diseases. However, most of the existing HDN studies are based on molecular information and can only partially describe disease interconnections. Building on the unique Taiwan National Health Insurance Research Database (NHIRD), in this study, we construct the epidemiological HDN (eHDN), where two diseases are concluded as interconnected if their observed probability of co-occurrence deviating that expected under independence. Advancing from the existing HDN, the eHDN can also accommodate nonmolecular connections and have more important practical implications. Building on the network construction, we examine important network properties such as connectivity, module, hub, and others and describe their temporal patterns. This study is among the first to systematically construct the eHDN and can have important implications for human disease research and health care and management.

Human disease cost network analysis

“…The wide connection of hypertension with a large number of diseases, for the Taiwan as well as other populations, has been reported in the literature, and is also reflected in our cost analysis. The overall top diseases share some similarity with the recent pHDN analysis . However, differences are also observed.…”

“…In the literature, multiple approaches are available for network construction. Here we adopt a weighted and undirected construction approach, which has been used in HDN analysis as well as genetic network analysis . Different construction approaches have their own advantages/disadvantages, and there is no dominatingly best approach.…”

“…Such studies have a limited scope and cannot reveal the “big picture” of disease interconnections. A major breakthrough is the HDN (human disease network) study, under which a large number of diseases are systematically analyzed. The HDN studies have the potential to reveal previously unknown disease interconnections and “globally” describe diseases.…”

“…Some studies use hospital‐ or community‐based data, and may have a sample selection bias problem . Some others are based on large insurance systems and use claim data . However, most insurance systems, such as Medicare and Kaiser, also have a sample selection bias problem and cannot describe disease properties for the general population.…”

“…6,11,15 Some others are based on large insurance systems and use claim data. 10 However, most insurance systems, such as Medicare and Kaiser, also have a sample selection bias problem and cannot describe disease properties for the general population. Some studies, for example, the renowned mHDN (molecular HDN), 16 are based on meta data extracted from publications.…”