Towards FAIR Data Sharing in the German Medical ......Towards FAIR Data Sharing in the German Medical Informatics Initiative Ganslandt T1, Sax U2, Semler SC3 1 Heinrich -Lanz Center

Towards FAIR Data Sharing in the German Medical Informatics Initiative

Ganslandt T1, Sax U2, Semler SC3

1 Heinrich-Lanz-Center for Digital Health (HLZ), Mannheim University Medicine, Ruprecht-Karls-University Heidelberg2 Department of Medical Informatics, University Medicine Göttingen3 Technology & Method Platform for Networked Medical Research (TMF) e.V. Berlin

Page 1

Grant 01ZZ1801E

The German Medical Informatics Initiative

MEDINFO 2019 Lyon - Workshop FAIR Data Sharing | Ganslandt T | Towards FAIR Data Sharing in the German MII | 26.08.2019 Page 2

Foster re-use ofroutine clinical data

Demonstrate utilitythrough clinical use cases

StrengthenMedical Informatics

as a discipline

150 M€ fundingby BMBF

Long-term perspective

The 4 MII Consortia


Image source: http://www.medizininformatik-initiative.de/en/node/5

http://www.medizininformatik-initiative.de/en/node/5

The Collaborative MII Governance Structure & Working Groups


Image source: https://www.medizininformatik-initiative.de/en/about-initiative/organisational-structure-and-actors

https://www.medizininformatik-initiative.de/en/about-initiative/organisational-structure-and-actors

Where can we realize FAIR along the research process?


Consent WG

Broad consent, Consent implementation

Data Sharing WG

SharedPolicies

HarmonizedProcesses

Central projectregistry

Interoperability WG

MII Coredataset

Metadataannotation

Research process

Interoperability WG Modular MII Core Dataset


Oncology Pathology findings

Imaging findings

PDMS/Biosignals

Biomaterial

Genetic tests Structure data

Billing codes

Cost data

Exte

nsi

on

mo

du

les

Diagnoses

Procedures

Lab findings

Medication

PersonDemographics Case data

Bas

ic m

od

ule

s

Definition of scope & priorities mandatory vs. optional modules based on data availability & relevance

to the consortial use cases

Abstraction of consortial data structures essential for cross-consortial data usage shared definition of data structures & terminologies based on HL7 FHIR profiles governance process, including HL7 balloting

FAIR

I1: broadly applicable knowledge representation I2: vocabularies that follow FAIR principles R1.3: meets domain-relevant community standards

Interoperability WG Metadata Annotation


Description of data availability “in general” regarding types & volume of data

(inventory & map of MII data, also curated sets) specifically regarding data protection restrictions

Description of data provenance “in general” regarding data sources, intention

of data capture and transformations specifically for core dataset modules (e.g. laboratory

equipment & test kits used to create a finding)

Description of data quality “in general” regarding overall quality metrics specifically regarding fitness for individual use

FAIR

F2: described with rich metadata R1: richly described with relevant attributes R1.2: associated with detailed provenance

Data Sharing WG Policies, Processes & Central Registry


Shared data usage policy including mandatory terms of service,

data transfer agreements & governance structures mandate to archive & share result data

Harmonized data sharing processes currently being modelled in BPMN to be supported by consortial software

implementations

Central project registry organizes cross-consortial access to data provides register of both consortial as well as

cross-consortial data projects

FAIR

F4: registered or indexed in a searchable resource A2: metadata stay accessible R1.1: clear and accessible usage license

Consent WG Capture & Application of Patient Consent


Establishment of broad consent document modular, including use of biospecimens,

insurance data and re-contacting the patients coordinated with state & federal data protection

offices as well as association of ethics committees

Implementation of consent structured electronic representation electronic capture of consent

FAIR R1: richly described with relevant attributes

R1.1: clear and accessible usage license

MII (upcoming) FAIR Achievements & Challenges


F2: described with rich metadata F4: registered or indexed in a searchable resource

A2: metadata stay accessible

I1: broadly applicable knowledge representation I2: vocabularies that follow FAIR principles

R1: richly described with relevant attributes R1.1: clear and accessible usage license R1.2: associated with detailed provenance R1.3: meets domain-relevant community standards

FAIR

No

t ye

tFA

IR F1: globally unique & persistent identifier F3: metadata includes identifier of referenced data

A1: retrievable by identifier over open protocol

I3: includes references to other (meta)data

local responsibilityfor archiving

limited detailsin central registry

availability/licensing ofterminologies (e.g. SNOMED CT)?

mapping to othercommon data models?

structured representation /compatibility to CC licenses?

Evaluation of Personal Health TrainIntegration of identifiers

into research process

Outlook: German National Research Data Infrastructure (NFDI)


NFDI 4 Medicine

FAIR Implementation for secondary use

as well as basic research

sustainability

current status:LOI received,

grant proposal TBD by October 2019

Towards FAIR Data Sharing in the German Medical ......Towards FAIR Data Sharing in the German Medical Informatics Initiative Ganslandt T1, Sax U2, Semler SC3 1 Heinrich -Lanz Center

Documents