CARDIOPROOF ‘s Bootstrapped Infrastructure

One of the central pillars of Cardioproof is the technical infrastructure that enables the safe hosting, transmission and analysis of the pa- tient data on which the project relies. Gnúbila, with consultation from clinical and technical partners, has put in place the first version of the infrastructure for the Cardioproof project. This infrastructure has been built on top on the ongoing FP7 project MD-Paedigree, as an extension to it, in order to make best use of time, effort and funding.

Installed on top of the existing MD-Paedigree infrastructure, the current solution provides two nodes that have the ability to host science gateways, central services and a web portal. The Cardioproof project shall add at least one node to this infrastructure. Most of the solution is provided by functionality and resource implemented for MD-Paedigree, bootstrapping on these existing components has provided a significant cost reduction in terms of hardware infrastructure, software evolution and maintenance. Both projects benefit from this arrangement, as any improvements coming from either project are shared with the other.

In Cardioproof, the data infrastructure and information system is based on gnúbila’s FedEHR product. As the BARC study “Big Data Survey Europe” from 2013 [1] states, organizations are taking a serious view on big data, recognizing the critical success factors and issues associated with handling enormous data volumes. Like medical imaging in the 80’s, big data is indeed about to reorganize medical practice. Big data not only is a major challenge for ICT and health care professionals, but also is a great opportunity. The use of massively available medical data may allow clinicians to simulate potential outcomes and so prevent patients from undergoing ineffective treatments or improve on current treatment. In other words, accumulating and using data to develop a greater understanding of pathophysiological processes will result in significant healthcare improvements.

Developed in collaboration with renowned medical centres in Europe, FedEHR is a patient-centric Electronic Health Records (EHR) big data solution. FedEHR stands for Federated Electronic Health Record. It leverages cloud elasticity to provide a scalable vendor- neutral database, which is able to cope with massive multi-modal and heterogeneous medical information, data and knowledge integration. FedEHR grew from leading edge technologies developed and tested in computationally and data intensive environments at the European Organization for Nuclear Research (CERN[2]). Over the last 7 years, FedEHR has matured and extended its range of application to a number of fields of medical science, from advanced biomedical research, to translational and clinical medicine. Fed- EHR has been installed in reference hospitals internationally, including Necker Enfants Malades in Paris, France, the Great Ormond Street Hospital in London, UK, Ospedale Bambin Gésu in Rome, Italy, and the Johns Hopkins University hospital in Baltimore, USA. The solution also was awarded at major events from its inception, including the Gold Medal at the International Inventions Exhibition in Geneva[3] in Switzerland, in May 2007; the Technology Transfer Award from CNRS[4]. in December 2010; and, as part of MD-Paedigree, the Best Exhibit Award of Europe’s largest ICT conference, ICT 2013[5], last November 2013.

infra3

infra2

Structural Overview

The system is composed of two main components:
  • a. A backend, composed by all the Web Services being the ground of the platform with the addition of external applications
  • b. A frontend, offering a user-friendly web inter- face allowing user communities to easily interact with the platform
Cardioproof’s backend has been inherited from MD-Paedigree and is the foundation of the system. Its Service Oriented Architecture (SOA) comprises a range of components interacting with each other and with external entities. Most of the platform features are exposed as Web Services, allowing an easily decoupling of the user interface rendering from the business logic. The user Interfaces can offer powerful features by aggregating the Web Services. The main entry point to the system is the frontend, and has been built by extending the MD-Paedigree front end. It exposes all the Cardioproof applications and tools through a common and simple web-based interface.
aa

Patient Centric Data Structure

In accordance with the latest data modelling concepts in the literature, Cardioproof implements a storage model that is fully centred on patients. All data that is stored in the system is organised around a data structure representing a patient model. The current description of FedEHR architecture provides an evolutionary structure of data starting from the patient. Currently, the data structure is oriented around medical concepts of medical events and clinical variables. These abstract models can be refined and specialised using metadata definitions created from physicians’ descriptions of diseases and exams or from normative works.

The New Patient Cart

The first need expressed during the project was to be able to easily choose a patient to add to a co- hort of study. To respond to this, the first version of Patient Cart portlet has been designed and implemented. This development is to add ability to select patients in all the existing interfaces that show, sort, or list patients on criteria, and to send these patients to the Patient Cart. In the future, this will allow users to generate, share, and save patient cohorts into the system.

The Query System

FedEHR provides an inter-site query system presented as an SQL query for end users. These queries are managed by a query management system, which generates a result set that can be downloaded in a variety of formats and also includes a graph visualisation tool. Stored data is not useful without a query system. Inherited from MD-Paedigree, the Query System has been updated to be able to enrich Cardioproof Patient Cart to add the management of patient lists.

Access Rights

The Cardioproof/MD-Paedigree system also includes a RBAC (Role Based Access Control) mechanism for data security. For each patient medical event, the access can be defined based on roles given or not to the users. The provider of the data decides what access is given to whom and is able at any moment to review and alter the access to data provided to a particular user. Roles can be given to groups that can be multiplied as needed. A group can be assigned to member of a whole hospital, a special service or represent inter-hospital group. to member of a whole hospital, a special service or represent inter-hospital group.

Data Importers

Data importers have bee developed to import data from routine systems, while adding metadata infor- mation from images and measurement and textual information from annotations. They also are able to normalise the data and pseudonymising it using a 3 part storage mechanism. In the process, some ontological information such as ICD or SNOMED can be added in order to provide a uniform concept-based query capacity. This means that there is a common human readable data structure at each node.

Our Current infrastructure

Basing the Cardioproof infrastructure on that already in place for MD-Paedigree has helped to jumpstart data collection by avoiding the delays incurred from waiting for additional hardware to be installed.

infra1

Cardioproof Access Rights Architecture

The Rome Gateway has been updated to allow Cardioproof data to be stored. The DHZB Gateway has been installed in Rome and is waiting for migration to the system in Berlin when available.

Next Steps

At least two further Nodes are planned for Cardioproof, one in London at UCL, and one in Berlin at DHZB. Once the hardware has been installed pre-configured virtual machines will be taken and installed on site. The added nodes mean that the system will have multiple physical nodes in three different countries, all collaborating to share data. As you can see, Cardioproof has benefited substantially from the existing infrastructure that has enabled us to achieve the current advanced status. With the proposed updates the Cardioproof’s infrastructure will go from strength to strength.

Authors: Sébastien Gaspard, David Manset, and Jérome Revillard

References:

[1] http://www.pmone.com/fileadmin/user_upload/doc/study/BARC_BIG_ DATA_SURVEY_EN_final.pdf

[2] Health Surveillance. How Knowledge Transfer Changed Biology, Medi- cine and Health Care. D. Manset. WILEY, 2014. In Press.

[3] Gold Medal at the International Inventions Exhibition in Geneva http:// cds.cern.ch/record/1035139 “MAAT Gknowledge took a gold medal for Mam- moGrid, a GRID-based mammogram analysis system to be implemented in the Extremadura region in Spain”.

[4] Technology Transfer Award from CNRS https://gnubila.fr/en_GB/awards- recognitions

[5] Best Exhibit Award at ICT 2013 http://ec.europa.eu/digital-agenda/en/ news/meet-winners-best-exhibitors-ict-2013 MD-Paedigree.