21-22 March 2019, Level Eleven – Amsterdam, the Netherlands
Data is one of the few technologies that becomes of greater value when it is shared. Patient related health data is no exception in this. Sharing health data is critical to spark data research and innovation, especially in the domain of the rare diseases. However, there are myriad challenges when it comes to making patient data findable, accessible, interoperable and reusable. During this meeting March 21-22 in Amsterdam, experts participated in a joint effort to enhance the sharing of patient-related outcome for data sharing in Duchenne.
Let us use our data
Prof. Dr. Barend Mons, former chair of the High Level Expert Group of the European Open Science Cloud starts with a surprising statement: Stop sharing data. Sharing data is associated with giving information away to a bigger entity that collects it. Instead, he insists on calling it visiting data, and data stewardship instead of management, a form of open science where the patient is co-pilot of their own health data.
Prof. Dr. Barend Mons, chair of the High Level Expert Group of the European Open Science Cloud
Moreover, a staggering 80% of accumulated health data is lost in 2 years, and 30% of healthcare costs are bad-data related. By FAIR-ification of data, and much better use of it, the health-related costs will decrease significantly. He requests a citizen plea to jointly exploit the value of patient data, and to think of tools on how to engage in distributed deep learning over data instead of data collection.
In 2016, he published the ‘FAIR Guiding Principles for scientific data management and stewardship’ in Nature’s Scientific Data. With this, he and his colleagues intend to provide guidelines to improve findability, accessibility, interoperability and reuse of digital assets. These principles emphasize machine-actionability; the capacity of computer systems to work with FAIR data with none or minimal human intervention. This because humans rely more and more on computer support (as some might call artificial intelligence) to deal with the increasing complexity of data.
Marco Roos, PHD assistant professor and involved in Elixir builds on the FAIR Principles that Mons introduced by stating “Don’t share, be FAIR”. Currently, he sees ‘data warehousing attempts’ which accumulate more and more data, that is then stored in siloes that no other party has access to. However, the only way to create value out of data is to share it with others and allow data to move around.
Marco Roos, PHD assistant professor and involved in Elixir
He explains there are two skills that are needed to use patient data to answer questions efficiently and correctly: domain specialists and data scientists. Domain specialists are patients, caregivers and clinicians that accumulate the data. Then, data scientists that are involved from the beginning to ensure data is collected in a FAIR way can then be simultaneously analysed, drastically decreasing the time to present new insights.
Furthermore, there are socio-technological and data compatibility challenges that need to be solved before there can be an attempt to scale this up. We need to think of a business model on how to ‘sell’ meta analysed patient data in a safe way to ensure health data will not get lost.
Personal Health Train
Prof. Peter-Bram ‘t Hoen, professor of bioinformatics at RadboudUMC states that FAIR compliance is the first step to building on existing Duchenne data. He refers to the FAIR Principles, an acronym for Findable, Accessible, Interoperable and Reusable data. The Personal Health Train is an ICT infrastructure based on the FAIR principles, which facilitates data visiting, not sharing, distributed health data and creates value for citizens, healthcare, and scientific research.
Registries, hospitals or personal lockers act as ‘stations’ which are interconnected by a ‘railway’ of interoperability specifications and standards, firmly based on a legal and ethical framework. The so-called ‘trains’ consist of algorithms that can then visit the data, perform analyses after the station ‘owner’ (a patient, patient organization, data cooperative, hospital, database owner) has given permission, and report the outcomes back.
Currently, we lack a legal framework when it comes on researching new data, whether it is sensitive (health data with medical confidentiality) or non-sensitive (data masking through pseudonymization or anonymization) patient data. Although the GDPR might seem like a impediment in breaking through the siloes of patient data, it changes the role of the patient and becomes a powerful privacy-by-design enabler when you include for example dynamic consent.
What we need, he concludes, is incentives for jointly realising this effort for patients, regulators, clinicians, researchers/scientists and the industry to ensure all data initiatives in the future are FAIR compliant. By doing so, all stakeholders are committed to work to FAIR-ification of patient data that is beneficial for all.
The future we envision
Julián Isla, Founder of Foundation29, member of the COMP at EMA and Data and Artificial Intelligence Resource Manager at Microsoft presents the ongoing shift from analogue to digital data collection to move the complexity of data from people to computers. This empowers people to use technology in a positive way and empowers regulators to make better decisions. In rare diseases, where prevalence is low and drug development is moving at slow rate, digital data has the potential to change design of clinical trials, potentially creating positive outcomes.
Julián Isla, Founder of Foundation29, member of the COMP at EMA and Data and Artificial Intelligence Resource Manager at Microsoft.
He advocates for a shift from an ‘Internet of Things’ to an ‘Internet of Patients’, where patients instead of devices are sending and receiving data, communicating and interacting with other patients over the world. This also closes the feedback loop in data feedback that captures information by sending it to all stakeholders, making sure every participant benefits from the effort.
In the discussion that followed the presentations, special attention was given to ‘FAIR awareness’: the act of engaging patients in the effort of visiting data, empowering citizen science. This correlates with training on advocacy on FAIR principles to share success stories and tangible outcomes, to create both long- and short-term goals that encourage data visiting.
The cost of making data FAIR can be included in funding principles, encouraging projects to work according to the FAIR Principles. Some organisations as the Dutch Parent Project are already working with these principles, having similar guidelines as open access projects. However, there is no business model yet behind how existing data can be made FAIR, and how stakeholders can be rewarded for FAIR-ifing health data. Data means power and thus value (whether monetary or intellectual). With distributed power, who takes responsibility for this and receives value?
Now that the mindset has shifted from quantity to quality, another point of discussion is how to map and assess the quality of data. This also touches base with the reusable component in the FAIR acronym. Who will start looking for how to map and assess existing data, e.g. in post-marketing registries, and how is this done?
In the following break-out session, participants divided into groups of expertise to create commitment by formulating actions on how to contribute to FAIR Duchenne data. Respectively, the following topics were created:
- FAIR Awareness – actively engage patients/caregivers in data collection
- FAIR Incentives – discover gains per stakeholder and encourage collaboration
- FAIR Assessment – framework to map and reuse existing data
- FAIR Advocacy – dissemination of FAIR principles
New and ongoing initiatives in Duchenne data collection
10 years of Duchenne Registry by PPMD
Pat Furlong, founder president of Parent Project Muscular Dystrophy, presents the insights and lessons learned of ten years of Duchenne Registry, a central hub that brings together those with living with Duchenne or Becker, along with their families and caregivers, to connect them with medical research, clinical care, clinical trials, and each other. At the same time, it acts as a resource for researchers and industries, allowing access to aggregate de-identified. The Duchenne Registry is the largest, most comprehensive registry for Duchenne and Becker. Currently, data from the Duchenne Registry will be connected with patient-report data, clinician reported data, electronic health records and post marketing surveillance data provided by industry partners with approved therapies in the result of DORI, the Duchenne Outcomes Research Interchange initiative.
TG-DOC by TREAT-NMD
Prof. Dr. Nathalie Goemans explains the TREAT-NMD Global Database Oversight Committee (TGDOC) that consists of representatives of the TREAT-NMD network and representatives of patient organisations and national registries. This initiative is responsible for reviewing requests for data from the global registries. This is intended to be a streamlined and rapid procedure in order not to delay approval. They aim to respond to requests within a set time period of two weeks. Currently the plan is piloting a new registry for Post Market Registry in the SMA community, and copy this concept for DMD however there are no outcomes defined yet for this initiative in the Duchenne field.
Duchenne Data Platform by Duchenne Parent Project Netherlands
Yolanda Ludeña, project manager for health projects at Foundation29 presented Duchenne Data Platform (DDP), a collaboration between Duchenne Parent Project NL and non-profit organization Foundation29 for the benefit of the wider Duchenne community. This data platform gives patients the power and control over the use of their own data and the option to ‘get their data together’. See it as ‘storing in a locker’. Patient data can be used for questions relevant to the patient community whether it is development of new drugs, new technologies or about daily life. It will not only make data available to return the data to patients and their parents/caregivers. DPP is a data platform for storing biomedical data as well as social data and has been developed to store data easily but in a highly secure environment.
The Duchenne Data platform will facilitate the adoption of emerging technologies regarding data collection and enable their optimal application in health research care and drug development. The platform is via the internet and on mobile devices to easily access personal health data for Duchenne patients and families.
Panel & brainstorm
The second and final day, the meeting started with the exploration of various perspectives on data sharing from patients, regulators, clinicians, preclinical researchers/scientists and the industry. This panel discussion, led by Marco Roos and Julian Isla resulted in 6 topics. These topics formed the basis of a brainstorm in working groups, which had the following outcomes:
- Next steps – start the FAIR conversation through hackathons, mapping datasets and create a Duchenne FAIR consortium
- Incentives for industry and regulators –e.g. market access, challenges and thresholds per stakeholder, economic incentives for reimbursement
- Manifesto – draft and publish manifesto with guidelines on visitable data and dynamic consent, get endorsement of all stakeholders
- Incentives for patients – find peers, ask questions related to care, quality of life and clinical trials, fostering innovation
- FAIR assessment – choose a standard, map options and talk to other rare diseases (‘cross FAIR-ification’), funding opportunities and assessing FAIR levels in existing initiatives
After the meeting, next steps are organised to make FAIR data for Duchenne happen to enable data visiting for the benefit of the patient. Soon you will hear more about the working groups about the Manifesto and the B.Y.O.D. (Bring Your Own Data) workshops.