=Paper=
{{Paper
|id=Vol-2759/keynote1
|storemode=property
|title=An Ecosystem for Precision Medicine: Closing the Cycle from Doctor to Patient
|pdfUrl=https://ceur-ws.org/Vol-2759/keynote1.pdf
|volume=Vol-2759
|authors=Irini Fundulaki,Tzanina Saveta,Vassilis Papakonstantinou,Yannis Roussakis
|dblpUrl=https://dblp.org/rec/conf/semweb/FundulakiSPR20
}}
==An Ecosystem for Precision Medicine: Closing the Cycle from Doctor to Patient==
https://ceur-ws.org/Vol-2759/keynote1.pdf
An ecosystem for Precision Medicine: Closing
the Cycle from Doctor to Patient?
Irini Fundulaki12 , Tzanina Saveta1 , Vassilis Papakonstantinou1 , and Yannis
Roussakis1
Institute of Computer Science-FORTH
fundul@ics.forth.gr
National Infrastructures for Research and Technology (GRnet)
fundul@admin.grnet.gr
1 Extended Abstract
According to the Precision Medicine Initiative1 , precision medicine is ”an emerg-
ing approach for disease treatment and prevention that takes into account indi-
vidual variability in genes, environment, and lifestyle for each person”. Precision
Medicine allows treating doctors to decide on the appropriate treatment for a
patient who is suffering from a specific disease and who has a specific genotype.
The Precision Medicine approach differs from the traditional approach in which
the treatment of diseases follows a ”one-size-fits-all” approach focusing on the
average person without considering the differences exhibited by the various pa-
tients. Precision medicine is a model in which the individuals phenotype and
genes, environment, and lifestyle are used as additional layers of patient data in
disease treatment and prevention plans.
All Precision Medicine strategies include the use of decision-making pro-
cesses based on the use of biomarkers. Biomarkers can be used to augment and
refine traditional medical data to provide a higher level of specificity for disease
prevention and patient treatment. The knowledge of the patient’s genotype is
crucial in providing the appropriate treatment to her. From this perspective the
field of ’omics’ has played a very important role in the development of precision
medicine. There are different types of ’-omics’ data: (i ) genomics that study
the genome, or sets of genes (panels), (ii) epigenomics that focus on genome-
wide characterization of reversible modifications of DNA or DNA-associated pro-
teins (iii) transcriptomics where RNA genome-wide levels are studied, both from
a qualitatively and quantitatively perspective (how much of each transcript is
expressed), (iv) proteomics used to quantify peptide abundance, modification,
and interaction ( v) metabolomics that simultaneously quantify multiple small
molecule types (vi) microbiomics that investigate all the microorganisms of a
given community together and (vii) radiogenomics (imaging genomics) that fo-
cus on the relationship between imaging phenotypes and genomics.
?
Supported by the National Network For Genetic Cardiovascular Diseases Study And
Prevention Of Sudden Death In The Young On The Basis Of Precision Medicine (GR
iCARDIACNET)
1
https: // obamawhitehouse. archives. gov/ precision-medicine
Copyright c 2020 for this paper by its authors. Use permitted under Creative Commons License
Attribution 4.0 International (CC BY 4.0)
2 Irini Fundulaki et al.
Genomics are used for diagnosis in many medical fields such as oncology,
cardiology, nephrology among others, and are explored in the field of neurology
and most importantly in the case of neurodegenerative diseases. Proteomics and
metabolomics, are now often incorporated into the everyday methodology of
medical researchers. The integration of different ’-omics’ data (multi-omics)
is used to elucidate potential causative changes that lead to disease, or the
treatment targets, that can be then tested in further molecular studies.
Producing solely ’-omics’ data and storing them in a repository does not suf-
fice for either clinical or research purposes. Such data are useful only when they
are associated with the patient’s clinical information. For example, in the case
of inherited cardiac diseases, the geneticist can determine whether a mutation is
pathogenic, non pathogenic, of unknown significance (VUS) only if she is aware
of the patient’s clinical data such as her symptoms, comorbidities, and family
history of heart disease or sudden death in the family or other related diseases. In
order to serve in an optimal way the researchers (clinicians, geneticists) it is ab-
solutely necessary to interconnect ’-omics’ data with the patient’s clinical data.
To support such interconnection, one needs to follow a patient-centric design for
the ecosystem of technologies and tools that will be used to serve the objectives
of Precision Medicine.
In Greece currently there are three initiatives for Precision Medicine: (a) The
Hellenic Precision Medicine Network on Cancer (August 2018 - August 2020,
extended to May 2021) (b) The National Network For Genetic Cardiovascular
Diseases Study And Prevention Of Sudden Death In The Young On The Basis
Of Precision Medicine (July 2019 - July 2021) (GR iCARDIACNET) and (c)
the National Research Network for Genetic Neurodegenerative Diseases (May
2020 - May 2022) (GR DeGeNEURONET). The ICT objectives of the last
two initiatives are:
– the creation of registries for the Inherited Diseases of the Heart (Cardiomy-
opathies, Aortopathies and Channelopathies) and Neurodegenerative Dis-
eases (Parkinson’s, Huntington Disease, Amyotrophic lateral sclerosis (ALS),
Alzheimer’s, Multiple Sclerosis) that store clinical patient data
– the creation of a Genomic Data Repository
– the creation of an Epidemiological Map that will lead to the identification of
populations with pathogenic mutations.
The utlimate goal of both networks is to use the data stored in the systems to
find populations who carry a specific genetic mutation and break the chain of the
inherited disease.
The design of all ICT systems is based on the requirements provided by
expert clinicians and geneticists who perform the samples’ analysis, as well as
epidemiologists who want to be aware of the epidemics of a disease in order to
decide on actions that must be undertaken in order to address the challenges
presented. To be able to integrate clinical and genomic data, we follow a patient-
centric approach in which each patient is uniquely identified by her social security
number known only to the treating doctor. The systems that we have built to
address the requirements of the two initiatives are the following:
Title Suppressed Due to Excessive Length 3
– a registry for inherited diseases of the heart for GR iCARDIACNET for
clinical patient data. The registries retrieve patient information from the
National Social Security Registry (ATLAS) and support the secure stotage,
and controlled access to sensitive health data.
– an e-prescription system that allows the treating physician to order genetic
tests for the patient
– an e-lab book system used by the lab personnel in the place of the traditional
lab book to support the complete life cycle of sample analysis. It supports
1. the sample recipient in recording the sample’s arrival date and its ap-
propriateness (b) the technical supervisor to assign the analysis to a
geneticist (c) the scientific supervisor to upload and sign the genetic
report and (d) the treating doctor to follow the progress of the analy-
sis. The system follows the ISO15189 standard that specifies the quality
management system requirements particular to medical laboratories.
2. a genomic data repository that stores for each analysed sample the
produced genomic primary data (FASTQ, BAM files), and secondary
data (VCF and annotated VCF files), the genetic report and metadata
thereof. The metadata support the efficient retrieval of primary and sec-
ondary data in the cases in which re-analysis is necessary.
The ecosystem of tools shown in Figure 1 is currently used by cardiolo-
gists to perform clinical diagnosis and research for hereditary diseases of the
heart in the context of GR iCARDIACNET. A cardiologist stores the clinical
data of the patient in the registry of inherited diseases of the heart (iCARDIAC
Registry). Once sufficient information is gathered, the treating physician can
prescribe genomic tests for the patient using the E-prescription system where
he chooses the appropriate panel (set of genes) associated with the patient’s
diagnosis. The prescription is then sent to the research lab, responsible for per-
forming the analysis along with the necessary patient’s clinical information so
that the lab’s scientific supervisor can provide the genetic report. Note that the
patient’s personal data (name, social security number, address, date of birth) is
not sent to the lab: the lab personnel know only the system’s internal identifier
provided for the patient. The scientific supervisor can then upload the primary
and secondary data produced by the analysis of the sample to the genomic
data repository along with their metadata and the genetic report which is
subsequently sent to the treating physician. The same ecosystem of tools will
be used in GR DeGeNEURONET(tailored for neurodegenerative diseases).
A more detailed presentation of the aforementioned systems can be found at
https: // sites. google. com/ view/ icardiacnet-services/ (in Greek).
In order to protect the sensitive health data we designed and implemented al-
ways in collaboration with the MDs, the Data Protection Officer of the Network
and the Legal Team, a role-based access control framework that implements
rules for roles (MD, Lab Personnel), group members (Member of Supervising
Committee, Scientific Committee, Genetic Tests Approval Committee, ... ) and
individuals. A user can be assigned more than one roles, and can be a mem-
ber of more than one groups. In addition to the access control framework, we
4 Irini Fundulaki et al.
anonymize the patient’s data: we assign to the patient a unique identifier used
in the complete data lifecycle that cannot be used to trace back the patient. For
instance, the treating doctor has access to all personal and clinical data of the
patient, but the lab personnel does not have access to personal data (i.e., she
has access to the patient identifier and to a subset of patient data that cannot
be used to identify her such as date of birth, place of origin).
Fig. 1. ICT Ecosystem for GR iCARDIACNET
In addition to the above, we installed all the systems in the secure infrastruc-
ture of GRnet (National Infrastructures for Research and Technology) to which
only authorized users have access. We use GRnet virtual infrastructure and we
have deployed the aforementioned systems on different virtual machines (see
Figure 2); data level encryption is also supported supported. Users are authen-
ticated with Keycloak rather than individual applications. Keycloak provides
Single sign-on (SSO) and single log-out (SLO) that allows a user to log in with
a single ID and password to any individual application and the user passwords
are encrypted using SHA512 encoding.
2 Building an ICT Health Ecosystem: Lessons Learned
The objective of an ICT Health Ecosystem is to create a holistic view of patients
data; in order to achieve this, we follow a patient-centric design that supports
the integration with National Registries (National Registry of Blood Donors,
Title Suppressed Due to Excessive Length 5
Fig. 2. Virtual Infrastructure of GRnet
Registry for the Chronically Ill, Registry for Diabetic Patients, . . . ), and e-health
systems (e-Prescription, Private Health Record, Hospital Information Systems).
Following such a patient centric design there are benefits to the patient who does
not need to carry her health record around (extremely important for people
with comorbidities), to the healthcare system since such integrated data can
be used for population genetics, for building epidemiological maps that can be
subsequently used for disease prevention and diagnosis, and to the researcher
(clinician, bioinformatician, geneticist) who has a tremendous amount of data
available.
There are different phases one should go through when building an ICT
health ecosystem. Gathering user requirements is not an easy task since MDs
are extremely hard to find and gathering their needs takes a significant amount
of time. The advice is not to wait for them, but to try and understand what they
need and propose them how their needs can be addressed by the systems that
will be developed. Furthermore, it is utopian to think that the ”waterfall model”
for system development would do: MDs want to see actual systems and use them
in order to understand if they address their needs. A report describing in detail
the system functionality will not do. Hence the advise here is to continuously
build prototypes and make them available to the users. Another important issue
that must be addressed is the issue of consent that dictates the extent to which
patient data (clinical and genomic data) can be accessed and reused and by
whom, inside and outside the context of the project, and for which purpose.
Consent should set out potential risks and benefits to participants, as well the
data anonymization procedures to be undertaken. The consent forms should
6 Irini Fundulaki et al.
be specified by the Data Protection Officer (DPO) and they Legal Team. To
do this, they should know the data flow and the processes implemented in the
ICT Ecosystem. A result of this effort is the specification of the access control
framework, i.e., who has access, and for each purpose and to what type of data.
This task is not easy since the MDs strongly believe that their patients’ data is
their data and should have total control over them (and unfortunately not only
they). According to the Greek Legislation, a treating doctor is obliged to share
for diagnostic purposes, their patients’ data. The consent form should specify
clearly the potential use of the patients’ data for research purposes by members
of the academic and research community.