Necessity for a novel form of interactive conference
Clinical trials are the foundation for the advancement of medical research, but they
are also complex, time consuming and expensive. To ensure scientific validity, clinical
trials must record large amounts of data on health and treatment of a carefully selected
group of patients or trial participants. That is why a robust and highly flexible
IT infrastructure that permits rapid reconfiguration to support different trials while
maintaining consistent data models is needed to ensure that clinical data can be shared
and used for analysis. Recently, EU projects, like TRANSFoRm, p-medicine, EHR4CR and
BioMedBridges, are creating advanced clinical research information systems consisting
of sets of tools that assist in the preparation and conduct of clinical trials and
the re-use of care data for research purposes. These tools are also able to assemble
data from heterogeneous sources to answer complex research questions, including functions
for data mining and support of research process workflows to meet the needs of translational
and personalized medical research [1].
Because the translation of biomedical discoveries into clinical applications is an
aim of the EU Commission, clinical research support is expected to be a major component
of EU funded projects [2]. In this context, the EU Framework Program 7 (FP7, http://ec.europa.eu/research/fp7/index_en.cfm)
has promoted translational research, the so-called ‘from bench-to-bedside’ approach,
that is expected to have practical benefits and improvements of the quality of life
for patients. In this context, it is planned that many of the tools developed in ESFRI
(European Strategy Forum on Research Infrastructures) and IMI projects will be used
by ECRIN, the European Clinical Research Infrastructures Network [3] to advance clinical
trials. To provide information on these new research tools and to evaluate their usability
for the clinical research community, the CRI Solutions Day was organised by ECRIN
together with other EU-funded projects (TRANSFoRm, EHR4CR, p-medicine, BioMedBridges
and ECRIN-IA) providing interactive “hands-on” sessions to allow visitors to see these
tools in action. The necessity for such a solutions day was based on the experience
of ECRIN that the conventional, class room style presentation of research tools is
not sufficient to demonstrate comparability and usability to many researchers. Thus,
presenters from academia, research infrastructures and EU projects presented their
developments with interactive sessions embedded in a framework of presentations and
joint discussions. The CRI solutions day took place at Heinrich-Heine University Düsseldorf,
in Düsseldorf, Germany, 26-27 May 2014.
A new landscape for modern clinical research
After the welcoming speech by the Rector of Heinrich-Heine University Michael Piper
(Heinrich-Heine University Düsseldorf), Christian Ohmann (KKS and ECRIN) opened the
conference with an overview on the current developments in clinical trials and the
influence that translational medicine and data-driven science may have on clinical
trials progress (Figure 1). Because still only very few innovations have managed to
get into clinical routine use, the need for change in drug development is evident.
It was proposed that data driven research will be able to discover structures in big
data and lead to predictions avoiding the detour through hypothesis-driven research
and clinical trials. One should ask, if it is possible to apply a unified modeling
approach to health care, and improve the effectiveness of health care with data-driven
medicine using for example genome-wide research data. But data-intensive science may
rather complement conventional methods and thus be used to increase potential benefits
following clinical trials to demonstrate efficacy and efficiency. For example, it
may be possible to use big data to identify subgroups of patients that will profit
from a treatment; or test oncological drugs to receive a better therapeutic prediction.
Whatever the case, the clinical trials landscape will change; EHR data will be used
for feasibility and patient recruitment. Increasingly predictive biomarkers will be
analysed to determine outcome, increase power and to stratify patients in clinical
trials.
Figure 1
Key notes speakers and EU project heads (from top to bottom: Christian Ohmann, Michael
Piper, Ann Martin, Brendan Delaney, Norbert Graf, Brecht Claerhout, Jacques Demotes,
Stephanie Suhr).
Ann Martin (IMI) explained the Innovative Medicines Initiative (IMI). IMI is Europe's
largest public-private initiative of collaborative research projects aiming to speed
up the development of better and safer medicines. About 2 billion Euros are being
spent with 50% of the sum paid by the EU and 50% by the pharma industry. It covers
until to date 11 calls for proposals resulting in 59 collaborative projects. About
130 Mio Euro is being spent only for improved knowledge management, covering integrated
biomedical data platforms and tools for data management and big data (e.g. eTOX, OpenPHACTS,
EMIF, eTRIKS). Two important aspects of all EU projects are aspects of descriptive
metadata and data quality / data governance. Interoperability must be established
by the implementation of standards. IMI II has begun with 3.45 billion Euro and employing
H2020 rules, with the first call starting about July 2014. Because IMI projects are
focusing on -omics (research including genomics, proteomics or metabolomics), genomics
and patient data, the platforms and tools discussed in this solutions day are of importance
for IMI projects, too.
Session 1: EU funded projects with participation of ECRIN
Wolfgang Kuchinke (Heinrich-Heine University, Düsseldorf) described the role ECRIN
plays in the EU projects TRANSFoRm, p-medicine, BioMedBridges and EHR4CR and how ECRIN
plans to employ tools developed in these projects. ECRIN participated in developing
information models, data models and ontologies, promoted the use of standards and
created legal and ethical frameworks for these projects. Most of the developed tools
will be provided as services and these services have to be integrated into the ECRIN
infrastructure of clinical trial support. This integration must proceed on three levels:
technical level, process level and business level. For ECRIN, the integration of data
stored in hospital information systems (HIS) and EHR for research purposes will be
a special challenge, requiring the collaboration of local hospitals, EHR and HIS vendors,
patients, investigators, ECRIN EU Correspondents and clinical trials units (CTU) staff.
Brendan Delaney (King’s College, London) explained the importance of the concept of
the Learning Health Care System (LHS) for the TRANSFoRm project [4]. Enabling the
use of primary care data for research, the final aim of TRANSFoRm is the improvement
of patient health and safety. TRANSFoRm consists of three parts: first, support of
epidemiological genotype-phenotype studies; second, support for randomized clinical
trials (RCT), and third, decision support for the physician. Pilot studies with the
tools developed in TRANSFoRm will begin soon. EHR data sources will be analysed using
the Query Workbench. The randomized clinical trial will employ an eCRF integrated
in an EHR that provides a trigger mechanism for indicating to the physician eligible
patients. To bridge different clinical concepts between the usage of clinical data
in care and research, ontologies were implemented in the tools (CDIM). In addition,
a full data provenance trail is ensured, and standards, terminologies and archetypes
are combined. To illustrate the recent developments of IT in clinical research; Brendan
Delaney used the picture of the evolution of the human brain. The legacy systems together
with associated terminologies, like HL7, CDISC ODM represent the reptilian brain;
the interactive and semantic aspects of IT solutions, like models and archetypes,
are represented by the limbic brain and the medical reasoning based on ontologies
is represented by the most evolved form, the neocortex.
Norbert Graf (USAAR) presented results and challenges of the p-medicine project [5],
a clinically driven project about personalized medicine in oncology with 19 partners.
The p-medicine platform serves knowledge discovery, including genetic and phenotype
data. It allows the assessment of analytical results in context to ensure that patients
obtain an individualized therapy. Thus patient empowerment must play an important
part in the project. Tools, models, means to ensure re-use of data, and data warehouses
are enclosed by a security framework, consisting of data anonymisation, contracts
with users and consents given by data donors. Clinicians will be the users of the
tools, but p-medicine will also ensure that more patients are enrolled in personalized
medicine trials.
Brecht Claerhout (Custodix) presented EHR4CR [6], an IMI project that will enable
the re-use of hospital data for optimising clinical trials conduct. The need for the
project results in the fact that drug innovation has become too slow and too expensive
(about 1.2 billion Euro is needed to develop a drug from a chemical entity). Most
important, 50% of delays in clinical trials come from inefficient patient recruitment,
because clinical trials often fail to enrol enough patients. EHR4CR has four use cases
for EHR data re-use: optimizing protocol feasibility, faster patient recruitment,
improvement of study conduct by pre-filling eCRFs and finally the capture of adverse
events (AE). The services developed will access real clinical trials information by
running queries at different clinical sites that compare if criteria are met. This
procedure will be able to speed up recruitment by giving physicians and investigators
means to search for eligible patients. The EHR4CR project is characterized by an open
architecture, a service platform, use of standards and is open to different semantic
layers. It has now reached the testing phase of its pilots.
Jacques Demotes (ECRIN) presented ECRIN-IA and concentrated on the project’s data
management aspects. ECRIN has become an ERIC (European Research Infrastructure Consortium),
a legal structure with now 5 countries involved, and 4 other countries that will join
soon. ECRIN-IA is about the structuring of user communities for nutritional research,
rare diseases and medical devices. Data management is part of ECRIN-IA and started
in the previous project with a large survey of all ECRIN centres [7]. This survey
showed that data management solutions for clinical trials are very heterogeneous and
differences in quality standards exist at data centres. What ECRIN needs is a system
easy to use in all ECRIN data centres that is cost effective to implement and extensible.
Addressing the need for quality standards in clinical data management, a certification
program for ECRIN data centres was developed covering GCP compliance and ECRIN requirements.
A suitable data management solution (VISTA) is being built by EORTC according to ECRIN
specifications. VISTA will be available at low cost for academic institutions and
for ECRIN. In addition, ECRIN plans to develop a clinical trials outcome database
that will store core data sets of outcome measures of ECRIN trials to enable further
meta-analysis after the end of the trials.
Stephanie Suhr (BioMedBridges) presented BioMedBridges, a large EU project in which
21 partners represent 10 new European research infrastructures covering research domains
from bioinformatics and biobanking to translational research. Recently, emerging infrastructures
for systems biology and microbiology have also joined. The aim of BioMedBridges is
to integrate existing islands of data sources that harbour clinical data, biological
data, and images in different research infrastructures by the means of data bridges
and services. With the help of REST services and a data security framework, data previously
isolated will become discoverable for researchers. Data integration will take place
on the technical level (standards) and on the process level by REST services and by
semantic web techniques. Three tools developed by BioMedBridges were presented at
the workshop, CTIM, MOLGENIS and XNAT. In addition, LAT was developed, a tool that
precedes the data discovery process and supports researchers in identifying legal
and ethical risks associated with the process of data access or data sharing.
The following joint discussion focused on the provision and the sustainability of
the multitude of tools developed in EU projects. Because these tools are public developments,
they should be freely shared over the web (Ann Martin). In this context all speakers
confirmed the future public availability of the tools developed. Stephanie Suhr affirmed
that all deliverables and tools developed in BMB will be public. Though, ECRIN deliverables
are public, VISTA will not be open source and usage will be accompanied by a non-profit
fee. For the p-medicine project, too, all tools will be Open Source (Norbert Graf).
All deliverables of TRANSFoRm are publicly available; and TRANSFoRm tools will be
provided as Open Source for free usage with accompanying documentation. EHR4CR too
will make deliverables publicly available.
A discussion followed about the role that a certification procedure can play in infrastructures
and for tool employment. A certification of tools could ensure the GCP compliance
of usage or the observance of regulations, like the ones for data protection. ECRIN
has developed a certification procedure with an independent certification board for
clinical data centres. During the phase of certification development the German Accreditation
Agency advised ECRIN not to start with establishing an accreditation, but to create
a certification process (Christian Ohmann). Norbert Graf explained that p-medicine
has launched a special legal entity to ensure its sustainability: STaRC. All tools
in p-medicine must be validated for GCP compliance and usability by clinicians. But
an important aspect is that the validation and certification of tools that will be
used in clinical trials is out of scope of EU projects. In addition, lawyers are needed
to set up contracts between providers, data owners and users. For BioMedBridges, Stephanie
Suhr explained, sustainability is built already into the project by the participating
research infrastructures which are themselves sustainable entities.
Conference participants noticed similarities in tool development. For example, TRANSFoRm
and EHR4CR have developed tools/services with several aspects in common, like the
search for patient’s inclusion and exclusion criteria. It was asked, if both projects
will not develop tools or deliverables together. But the requirements underlying these
projects are quite different. Brendan Delaney explained that for TRANSFoRm the inclusion
/ exclusion criteria model is based specifically on PCROM and BRIDG, and both models
determine how the query workbench can operate and how the queries are created. Design
of queries is based on LexEVS terminology [8], and is linked over data node connectors
with ontologies. TRANSFoRm is a distributed infrastructure and allows that data stays
where it is. In contrast, according to Brecht Claerhout, the EHR4CR project starts
from a clinical model, ETL to the data warehouse. All queries are built upon a specific
information model (blue model and ECLECTIC) and the concepts of the information model
are again mapped to queries.
Break out session 1
In this breakout session (Figure 2), tools for clinical data management (OpenClinica,
ObTiMA, VISTA), tools for bridging experimental and clinical research data (tranSMART,
MOLGENIS, i2b2), and tools to integrate EHR data for clinical research (Feasibility
Service, Patient Screening Tool, Recruitment and Feasibility Tools) were presented
and discussed.
Figure 2
Impressions from the break-out and software demonstration sessions.
Break out session 2
In the second breakout session (Figure 2), additional tools for clinical data management
(Functional eCRF, mobile eHealth Solution), imaging tools for clinical research (XNAT
imaging pipeline, DoctorEye), and tools for biobanking in clinical research (p-BioSPRE,
Biobanking Catalogue, BBMRI Catalogue) were presented and discussed.
Panel discussion: what to do with all these tools?
In the previous breakout sessions important new tools to move clinical research towards
clinical trials in translational and personalised medicine were presented. The overall
impression is that the process of simple data collection has become insufficient for
clinical research and study feasibility and patient recruitment has to be supported;
EHR data, mobile eHealth data, biobank data and data warehousing solutions have all
to complement the data collection process. Töresin Karakoyun pointed out that for
clinical research and data collection the ability for cooperation has become of decisive
importance. Thus cooperation between EU projects and between research infrastructures
should be increased. While projects and infrastructures are often aware of the existence
of tools developed elsewhere, improved cooperation capabilities should enable the
exchange of resources and people as carriers of expertise between projects and infrastructures.
In this way, researchers will be enabled to find the best solution and the best tool
for their research. Users of tools and tool developers should learn more efficiently
from each other. The ESFRI infrastructures may work better together through BioMedBridges.
As a first step, BioMedBridges will create a repository of all tools, documents, sources
of the member research infrastructures.
The validation and certification as a necessary condition for sustainability is far
from easy requiring the creation of a large amount of documents for system validation,
like requirements specifications, functional specifications, and user specifications.
This leads to the issue that in the academic world the process of software development
is not as clean and precise as in industry. For example, tool development is often
an integrative part of a research project, as a result of which requirements become
a moving target. Thus, often no frozen version of a tool is available for its validation.
Once one has decided to freeze tool development for the validation / certification,
research still goes on resulting in new requirements and changes in tool development.
Norbert Graf clarified the need for certification of the tools developed in academia.
The ability for certification depends also on the way projects and their developed
tools are funded. As a consequence, the EU should provide funds to help projects to
move into sustainability. But to provide tools as Open Source is not an easy solution
either (Christian Ohmann); because Open Source, too, needs support, services, and
regular updates, requiring somebody who pays for the sustainability of Open Source.
We are now at a point, where the gap between pure research focusing on new problems
and publications and business focusing on the generation of money has to be bridged
for the EU projects. Projects must now be helped to move from a research approach
to a product development approach.
Jacques Demotes turned the discussion to the user communities for the developed tools.
IT experts are concerned that tool development is not driven by the user communities.
For example, clinical trials representatives have specific needs but cannot develop
tools by themselves. Thus, as a first step priorities and strategies should be defined
for the clinical trials communities. Although user requirements for clinical data
management systems (CDMS) exist in ECRIN, they cover only the basic aspects of data
management. But CDMS requirements should consider new developments in clinical trials,
like the involvement of imaging, biobanking and genome data. Here new tools are necessary
and must be integrated in existing clinical data management processes. In addition,
tools that are employed for clinical trials must be system validated for GCP compliance,
requiring specifications for the tools that cover these new fields of use. As a consequence,
ECRIN standard requirements that have been developed for the certification of ECRIN
data centres should be extended accordingly. In addition, the patient point of view
must be considered for clinical trials more strongly; patients will tell us more and
will give us more data. On the other hand, patients should be able to decide how their
data and biosamples should be accessed and used, and how the right informed consent
is employed.
Conference participants recognised a competitive situation between the tools shown;
for example, the tools ObTiMA, OpenClinica, VISTA support data management in clinical
trials, whereas tranSMART and MOLGENIS are data warehouse based solutions. It was
suggested to increase the transparency of developments and improve the exchange of
tools, information, as well as source code created in different projects. Norbert
Graf recommended building tools in a modular fashion, like it is done in p-medicine,
and with standardized interfaces. Ontologies should be able to link directly to items.
In this way, researchers will be able to select and assemble appropriate modules of
different tools for their research question. Often not everybody needs all functionality
available. This idea found support in the conference. Tools may come and go, but the
focus of software development should be on standardized interfaces and data standards.
One important standard in this context is IHE (Integrating the Healthcare Enterprise),
providing specific profiles for data retrieval. Standards are able to bring different
tools together, like in the case of CDISC ODM, a data exchange standard that can be
used to integrate different CDMS to be used in single clinical trial [9]. Töresin
Karakoyun pointed out that the area of processes should not be forgotten, because
processes are often still far from being clearly defined, like it is the case of the
workflow for personalised clinical trials.
ECRIN and data in multinational clinical trials
Jacques Demotes (ECRIN) was looking forward to identifying the challenges that are
encountered by clinical trials infrastructures. For clinical trials addressing innovative
health products, exploration of new indications for existing drugs, comparative assessments
of efficiency and safety of approved healthcare strategies, new infrastructure solutions
are necessary that need collaboration between industry-funded and non-industry funded
approaches. It is the lack of funding and the missing infrastructures that are the
reason why only 3% of non-industry funded trials are international. ECRIN-ERIC is
a combination of national networks of CTUs with national organizational points by
EUCos (ECRIN European Correspondents). Employing this infrastructure, ECRIN provides
all kind of services, like assessment of feasibility, logistic, contract development,
coordination with ethics committees, site monitoring, etc. IT should enable researchers
to better use health data for clinical trials, avoiding any duplication in data collection.
But only a small number of patients is actually being enrolled to participate in clinical
trials or to form part of a cohort study. The systematic use of health data for research,
including the integration of registries, cohorts, and clinical trials nested in cohorts,
international cooperation in non-commercial clinical trials must be facilitated by
establishing national / regional / global networks of co-operation in clinical science
accompanied by increased patient involvement in the clinical trial process. To support
these objectives, ECRIN expects that IT should improve the access to clinical data
(supporting data transparency and making patient-level data available for meta-analysis),
enable up-to-data analysis of all different types of clinical data (the ability of
meta-analysis should be mandatory), provide suitable anonymised patient data for analysis
(for example, retrieval of aggregated data sets, support of interoperability by using
standardized outcome measures, and by improving the integration of CRF data with high
throughput patient data in clinical trials (genomics, proteomics data).
In the subsequent discussion growing concerns in the research infrastructure community
were addressed that have to do with the fact that research conduct is changing from
research done by small teams of post docs to large anonymous teams in big projects.
As a consequence, collaboration, contribution and the authoring rules have become
complex and often unclear; research contribution cannot be based solely on authorship
anymore.
A first step to increase transparency in clinical research is to make the data collected
by CRFs public, not only the summaries of results and eligibility criteria. But this
must be followed by providing access to the study protocol, and additional data like
trial performance data. Access to all clinical trials data requires sophisticated
access policies, to prevent any misuse of data (e.g. the prevention to identify research
participants), as well as data provenance and a complete audit trail.
Session 2: translational research projects and their tools for clinical research
John Overington (EU Openscreen) presented the EU Openscreen infrastructure that combines
chemistry with biology and consists of national content collection centres. EU Openscreen
is a distributed infrastructure enabling trans-national access for the development
of bioactive small molecules. All databases are open access; and additional screening
centres with high-end equipment exist. Recently, an ERIC for EU Openscreen is being
negotiated.
EMIF was introduced by Alvaro Cortes (EMIF). EMIF is an IMI project with the aim to
develop a common information framework of patient-level data to facilitate access
to diverse medical and research data sources. It started in January 2013 with 48 partners.
Because it provides access to individual data sources and tries to enhance already
existing data sources, EMIF can be seen as a federation of databases; whereby each
data source retains full control over its data. Database owners are supported by the
provision of means for anonymisation / pseudonymisation, tools for linking ontologies
and by support of semantic interoperability.
Anka Bucur (Phillips) introduced the INTEGRATE project dealing with integrative cancer
research with the aim of facilitating clinical research to improve the treatment of
cancer patients. INTEGRATE focuses on using knowledge sharing for patient recruitment.
Recent use cases are in breast cancer, but the solution developed will be applicable
to all disease domains by providing a semantic interoperability platform. Tools for
patient screening and pathology review were shown in the hands-on sessions during
this workshop. An additional tool deals with cohort selection by providing means for
filtering data, querying databases, and visualization of selected cohorts.
Jan-Willem Boiten (TraIT) presented issues of translational research. The TraIT project
develops an IT infrastructure for translational research with the aim to facilitate
the collection, storage, analysis, archiving, and sharing of data by providing means
to connect clinical phenotype data with disease biology data. It started in 2011 as
a Dutch national initiative between CTMM, the Dutch Cancer Society, the Dutch Heart
Foundation, the Netherlands Bioinformatics Centre (NBIC), the Parelsnoer Institute
(PSI) and others and developed in the meantime to a consortium that consists now of
27 partners. The project uses already available tools that are integrated into an
IT infrastructure. TraIT will demonstrate the tools OpenClinica, XNAT, BBMRI catalogue
and TranSMART in hands-on sessions during this workshop. Already 80 studies with about
700 users are conducted with OpenClinica as data management system for clinical trials.
Peter Wittenburg (EUDAT and RDA) introduced EUDAT, a project that provides common
data services for the entire research community. EUDAT is a collaborative data infrastructure
consisting of a federation of trusted centres with the focus that its community people
are driving the project. The services available are B2FIND for finding research data,
B2SAFE for safely replicating research data, B2SHARE for storing and sharing research
data with added semantics and B2STAGE for preparing data for high performance computing.
An important aspect of EUDAT is the Persistent Identifier (PID) service. In principle,
every bit of data may get a PID as a kind of tag when it enters EUDAT. Another tool
is the AAI service offering secure access to scientific systems. Recently, EUDAT collaboration
projects have started, with about 30 submissions for participation. ECRIN plans the
testing of policy mechanisms for sensitive data together with EUDAT. Because big national
centres are the actors in EUDAT (e.g. CSC, SARA, RZG) the sustainability of services
is ensured by agreements.
Niklas Blomberg (EBI) spoke about ELIXIR. ELIXIR will provide necessary facilities
for researchers providing a multitude of tools required by academics from bench biologists
to chemo-informaticians. In this way, it supports work with the rapidly growing amount
of information about living systems. The ELIXIR infrastructure includes databases
(e.g. ENSEMBL), reference data, tools for data access, tools for the exchange and
analysis of sensitive data, standards for data exchange, encoding and integration
of data as well as training for researchers. To maintain core data resources, sustainability
plays a role in the pilots of ELIXIR. The EU genotype / phenotype archive, which contains
individual genomes, has raised the problem of identifiability of individuals from
their genetic information. As a member of the BioMedBridges project, ELIXIR helps
with a tool for the consistent identification of biosamples (BioSD) that allows finding
and linking 2.8 Mio samples in 32 sources worldwide.
IT challenges for innovative clinical trials
The keynote speech was held by Norbert Graf (USAAR) about the topic of IT and innovation
in clinical trials. In areas outside clinical trials, IT development has proceeded
to create eBooks and eBanking. In the clinical trials area such a development is still
missing. But as an example what is already possible, paediatric oncology can be seen
as a success story, where nearly 100% of patients are able to participate in a clinical
trial and where treatments show striking successes. On the other hand, still only
5% of adults are enrolled in prospective clinical trials. In addition, in paediatric
oncology, personalized medicine has become very successful. But personalized medicine
is rather complex requiring many different kinds of data, even data from system biology
models. The important point is that all necessary patient data is available for the
physician in due time to be used for clinical decision support. Thus, tool development
in personalized medicine must be clinically driven, evaluated, validated and in a
standardised form to allow the reuse of data. In addition, personal data must be secured
by a security framework (that once being developed can be re-used for other projects).
The use of common ontologies and semantic mechanisms allow data sharing and integration
and may be shared between projects, too.
The principal investigator may be supported by tools offering electronic informed
consents. But it has to be considered that often a global informed consent is not
possible, for example in research with biomarkers. In this case, the investigator
has to re-contact the patient to receive a new right to use biosamples and data for
future research. This situation illustrates the apparent value conflict that exists
between the individual treatment (privacy protection) and the collective interest
(access to data and research with data). Here a balance has to be found between both
exclusive interests without compromising the transparency for the patient. In this
respect, recent approaches including the ones of PatientsLikeMe, where patients can
upload their clinical data, but where the personal clinical data belong to the service
provider, who is allowed to sell them for example to insurance companies, should be
avoided. This is not the form of transparency that will support and improve research.
What is often forgotten is that innovative clinical trials demand not only novel IT
support, but a higher degree of teamwork between different stakeholders, like infrastructure
providers, legal/ethical people, tool users, trainers, validation experts, and patients
as one additional partner. Therefore, to concentrate research only on data-driven
approaches will not improve research, but will result in researchers being lost in
data space.
Returning to the p-medicine project, data management is done with ObTiMA, a tool that
has been demonstrated in this workshop. The experience of the project with software
development is that in general, tools should be built in modules, to allow combination
of different modules to adapt to changing requirements, and to ease interoperability
between tools. The integration challenge for personalised medicine data consists in
the necessity that patient data from different hospital information systems (HIS)
and different push/pull specifications must be considered and joined. Here, integration
can be simplified by putting HIS data into a standardised data warehouse. While in
the research area it is often possible to use different tools that all are able to
access data of patient cohorts or anonymised medical data, in the area of decision
support only the same tool should be used to ensure uniformity in decision support.
Decision support tools must also be able to access personal data of a single patient
under conditions of real time constraints. This gap between basic research and clinical
application can only be closed by interdisciplinary teams.
Break out session 3
In the third demonstration session, tools for clinical research support (Clinical
Trial Information Mediator, Pathology Reviewer, PASTEL) and EHR associated tools for
clinical data (Query Workbench, Patient recruitment service) were presented and discussed.
Final joint discussion: conclusions and how can we achieve sustainability of tools
and services?
The first CRI Solutions Day provided an opportunity to hear from a diverse group of
professionals working in research infrastructure management, clinical trials and biomedical
research about how they cope with the challenges of new developments in the field.
The format of the workshop consisting of lectures and hands-on sessions of tools with
developers was very positively received; it was appreciated that the focus was on
the developed tools and it was suggested to conduct such a workshop regularly each
year.
The first impression was that EU projects seem to cope with many similar problems
and have come up with similar solutions. For example, all domains struggle with the
same rigid conditions for data protection and the challenge of semantic interoperability.
Christian Ohmann indicated that several tools address the same problem, for example
four projects developed a query generator. But, do researchers need this kind of competition?
If in such cases a generic problem exists, it might be better to bring projects together,
work jointly on software development and increase transparency. Project funding too,
should consider this aspect in EU projects. Peter Wittenburg suggested that researchers
in different domains may be able to use the same tools for their research. In principle,
this problem is also a matter of the funding organisations and the way projects are
supported. On the other hand competition between projects and the development of similar
solutions may also be a good feature. Norbert Graf argued that it should be considered
that tools developed in EU projects often grow out of special research questions and
are based on domain specific requirements. Therefore, project members should compare
and examine what things that can be done together and what can be shared (e.g. common
ontologies used in different disease settings). Ann Martin agreed about differences
in the fundaments of EU projects and was surprised how different the projects actually
are. Thus, the observed similarity of tools may be only a resemblance on the surface,
but may hide deeper underlying differences in approaches and requirements. Töresin
Karakoyun took up this remark to address the problem of sustainability of tools developed
in EU projects suggesting that the next CRI Solutions Day should be about how to sustain
tools and provide support after the end of a project. It was suggested that industry
should be more involved, perhaps even connections with venture capital may be helpful
to sustain developments. But the main problem faced by all research projects that
develop tools is the need to shift tool development from the research area to a setting
of professional service provision. Jacques Demotes stated that sustainability should
be an outcome of a project. Limited ERIC funds are available to promote a couple of
tools that could be part of a clinical trials tool repository. But in the end, the
provision of tools must generate some revenue to invest in maintenance and further
developments. The infrastructures are learning to deal with the different challenges;
sustainability is one of these challenges and the many query tools may be part of
the game (Peter Wittenburg). Thus, projects should create and improve their business
models, work together with stakeholders and funders and tackle the need for seed money.
Though, as was suggested, infrastructures and projects may share components and tools;
it may need 2-3 years for a project to settle a joint understanding to be able to
share.
It followed a discussion of what the business models of the different tools presented
in the break-out sessions were. Several of the tools provide access to patient data
and thus, supply pharma industry demands. Industry laments the missing access to patient
data; negotiations with hospitals about patient data often last years. The problem
with hospital data and patient data in general is the restriction of regulations and
rules on the use of sensitive data for research purposes. In the area of patient care
and hospitals, trust in the industry does not exist; and the main concern of data
protection officers at hospitals is how any misuse of data can be absolutely prevented
(Christian Ohmann); an aspect generally overlooked by industry. To generate this necessary
trust, several aspects must become transparent: the reasons for access to patient
data, who is doing what with the data, and who is responsible for the data. In the
US, insurance companies are already selling patient data to defined questions from
industry. In addition, as a consequence of regulations and rules to protect data privacy
the developed tools must incorporate means for audit trails and provenance control.
In this context, the TRANSFoRm project has developed a comprehensive framework to
monitor provenance [10]. Nonetheless, to perform effective health research, systems
in hospitals must open. Patients too are waiting for treatment and are interested
in new research findings and therefore to a large degree support research with their
data. In such a situation a robust legal framework is needed to ensure that privacy
is protected and that non-anonymised data as well as genetic data is not shared in
an unrestricted way. In fact, laws that forbid physicians to share such patient data
already exist. An extension of the problem of data confidentiality is that tool development
in EU projects should not omit the patient empowerment perspective (Norbert Graf).
In the end data belong to the patient, and often the patient wants not only to share
data but also wants access to data. Thus, both aspects together, transparency and
trust, are of utmost importance for the development of tools. Perhaps one should show
these problems to policy makers, to find a way how to overcome them. Ann Martin pointed
out that in fact policy makers have taken up this topic and are willing to protect
patients; IMI is working on guidelines about secondary use of treatment data for research.
The final comment, the take home lesson, was stated by Töresin Karakoyun: 11 EU /
IMI projects, 7 research infrastructures, 22 tools and 92 participants came together
and possibilities for collaboration turned out to be a focal point for all projects.
Collaboration between research infrastructures and projects can proceed via a 7-stage
model moving from simple coexistence and communication to more intensive cooperation,
coordination and finally to coadunation [11]. At the moment EU projects are at the
point of cooperation and coalition building, but have not yet reached the possibilities
of true collaboration. To support active collaboration a kind of platform is needed,
to enable collaboration by providing help to decide how to organize it, how to pay
for it, how to support and maintain it. There should be rules and guides and a budget
for collaboration; technology alone is not enough.
Conclusions
The first CRI Solutions Day provided a unique opportunity to see, discuss and evaluate
the new tools that soon will change the way research is done, demonstrated by the
leading research infrastructures and EU projects in the field of biomedical research.
It became clear that still many challenges exists, especially the sharing and joining
of tools, collaboration and sustainability issues. Nonetheless, using new tools that
will provide advanced IT support, biomedical research and clinical trials will become
more interesting in future.
Financial & competing interests disclosure
The authors have no relevant affiliations or financial involvement with any organization
or entity with a financial interest in or financial conflict with the subject matter
or materials discussed in the manuscript. This includes employment, consultancies,
honoraria, stock ownership or options, expert testimony, grants or patents received
or pending, or royalties.
Author’s contributions
WK is the main author and collected the abstracts; WK, TK, SG and CO contributed to
the text and corrected the text.
Abbreviations
AAI: Authentication and Authorization Infrastructure; BBMRI: Biobanking and Biomolecular
Resources Research Infrastructure; BRIDG: Biomedical Research Integrated Domain Group;
CDIM: Clinical Data Integration Model (TRANSFoRm project); CDISC: Clinical Data Interchange
Standards Consortium; CRI: Clinical Research Informatics; CSC: IT Center for Science;
CT: Clinical Trial; CTIM: Clinical Trial Information Model; CTMM: Center for Translational
Molecular Medicine; CTU: Clinical Trial Unit; eCRF: electronic Case report Form; ECRIN:
European Clinical Research Infrastructures Network; ECRIN-IA: European Clinical Research
Infrastructures Network-Integrating Activity; EHR/EMR: Electronic Health Record /
Electronic Medical Record; EHR4CR: Electronic Health Records for Clinical Research;
EORTC: European Organisation for Research and Treatment of Cancer); ESFRI: European
Strategy Forum on Research Infrastructures; ETL: Extract, Transform, Load - process;
EUCos: ECRIN European Correspondents; EUDAT: European Collaborative Data Infrastructure;
FP7: EU Framework Program 7; GCP: Good Clinical Practice; HL7: Health Level 7; i2b2:
Informatics for Integrating Biology and the Bedside; IHE: Integrating the Healthcare
Enterprise; IMI: Innovative Medicines Initiative; INTEGRATE: Integrative Cancer Research;
IT: Information Technology; KKS: Coordination Centre for Clinical Trials; LAT: Legal
Assessment Tool; LexEVS: EVS terminology server; ObTiMA: Ontology Based Trial Management
for ACGT; ODM: Operational data Model; p-BioSPRE: p-medicine Biospecimen Search and
Project Request Engine; PCROM: Primary Care Research Object Model; PID: Persistant
Indentifier; REST: REpresentational State Transfer; RZG: Rechenzentrum Garching; SARA:
Stichting Academisch Rekencentrum Amsterdam; STaRC: Study, Trial and Research Centre;
TraIT: Translational research IT; TRANSFoRm: Translational Research and Patient Safety
in Europe; USAAR: University of the Saarland; XNAT: Extensible Neuroimaging Archive
Toolkit