mdm2neo4j - Medical Data Models to Graph Database

About the project

In medicine, large amounts of data are collected and generated every day. To advance patient care, it is necessary to analyse the data at hand. This however is only possible if data and metadata are findable. The Portal for Medical Data Models (MDM portal) provides over 25'000 reusable medical metadata forms in the CDISC Operational Data Model (ODM) format. Numerous structural metadata sets are available for reuse in a standardised format. However, exploring the growing collection of data forms remains a challenge. In this project, we aim to enhance the exploration options for the contents stored in the MDM portal by representing ODM as a graph.

In collaboration with the Institute of Medical Informatics at the Heidelberg University Hospital, we develop mdm2neo4j, a software tool that transforms metadata from the MDM portal into a neo4j graph. Thereby, graph-based exploration of the metadata is enabled together with complex queries across linked metadata sets and efficient syntax checking. ODM metadata becomes easily explorable, findable and retrievable. The mdm2neo4j is one example of how we utilise graph databases in our research.

About the software

The mdm2neo4j program extracts and transforms contents from ODM files into a labelled property graph and loads it into a neo4j graph database. The current version is available here. The code is published and can be reused under the GPLv3+ licence.