Healthcare ecosystems are siloed. This lack of interoperability & standardization makes it difficult to integrate personal health data (e.g. wearables, apps) with clinical data (either real-time accessible via connectors or persisted in e.g. Electronics Health Records (EHR)), access and jointly analyze it to generate insights. These siloed ecosystems are even more challenging for the increasingly common remote monitoring & digital therapeutics tools that gather real-world data & insights. E.g. Byteflies (BY) enables the follow-up of +4000 patients with chronic illnesses across >30% of Belgian hospitals and participants in pharmaceutical studies by recording medical data in the BY Cloud. The tele-treatment solutions (app with sensors) of MoveUP (MU) are rolled out for +5000 bariatric & orthopedic surgery patients. FAQIR Institute (FI) integrates health data (clinical, PROMS questionnaires) of +250 patients with a chronic rare disease, e.g. cystic fibrosis, for follow-up & clinical study. The siloed ecosystems create a barrier to leverage the data generated by these tools to achieve a holistic & longitudinal view on the patient’s condition, improve diagnosis, treatment & empowerment, and allow secondary use for research, innovation & policy making. This is especially challenging for patients with chronic or complex conditions who receive care from multiple providers.
The ownership, control, transparency & consent of health data use & insight generation are outlined in data processing agreements between patients, care providers & third-party companies. AContrario Law (ACL) (specialized in IP, IT, health, data protection & security law) experiences this process as complex, time-consuming & lacking patient involvement in data control decisions. This raises ethical concerns. Patients may not fully understand the implications of granting access to their data to third-party companies, leading to mistrust & reluctance to participate in data (re-)use projects. EU legislation will require that citizens can control their health data & ensure data portability.
As a reaction, a general shift is occurring to decentralized storage of personal data in which individuals store their data in personal data vaults & are able to decide at every moment who has access to specific data. In Flanders, this shift is already becoming a reality as the Flemish Government founded a data utility company that will provide citizens with data vaults & the FAQIR Foundation (FF) that develops a decentralized ecosystem backbone for healthcare.
For data analytics, this shift requires processing a very large number of small, individually permissioned data sets with varied formats & data models. Therefore, data standardization & access protocols across vaults are key, which is why the Flemish government & FAQIR are adopting the Solid standard to realize this decentralized & privacy-by-design data vision. Solid provides a collection of standards & data vocabularies based on Semantic Web technologies to set up data vaults that store data as documents containing Linked Data with the appropriate access control mechanisms, e.g. identity, authentication, permission lists. Linked Data is a method for publishing structured data using ontologies that can be linked together & interpreted by machines. Ontologies formally describe a all the concepts within a domain, their relationships & properties. Linked Data makes the semantics & context of the data clear & enables interoperability of the variety of data across vaults. Relying on Linked Data also makes it easier to adhere to FAIR principles: data must be Findable, Accessible, Interoperable & Re-usable.
However, current decentralized platforms, including Solid, do not meet healthcare demands to deal with complex & multi-modal data (e.g. longitudinal & densely linked data, detailed image data, sensor streams with high velocity) that need to be stored in an efficient & robust fashion and queried & analyzed in scalable way to provide insights. These solutions need to deal with complex permission & consent protocols that ensure patient privacy, while remaining scalable to foresee feedback to patients & clinicians in due time. There is a need to clearly differentiate between informed consent mechanisms for the use of health data for medical purposes and data protection consent mechanisms for the processing of health data for secondary use.
Within the PACSOI project, we focus on a specific use case, namely streamlining the health data flows for patients that undergo bariatric surgery as a treatment for obesity. We want to streamline this information flow both for the patients and their healthcare providers to let them achieve more insight in the patient journey to steer follow-up & treatment. We also want to support policy makers and researchers who aim to achieve high-level insights across this population to either train new data analytics techniques to improve the patient journey (e.g. optimal treatment suggestions) or steer policies. Pseudonymized data from a limited number of obese patients will be provided.
The student can choose to tackle on one or more of the challenges highlighted above, according to their interests:
- Investigate methods to integrate and store all the heterogeneous health data collected during a patient journey in a patient solid pod in an efficient and user-friendly manner
- Investigate methods to indicate in a fine-grained manner which parties have permission to process which types of data from each patient
- Investigate methods to process the huge amount of patient data in an efficient and/or trustworthy manner to gain insight across a patient journey or even across the complete patient population
- Design intuitive user interfaces to interact with this ecosystem, i.e. onboarding a new patient with all their legacy health data, configuring permissions, dashboards to visualize data, etc.
If the student has an interest within this domain that is not listed above, we are always open to new avenues of investigation!