Towards a Secure Processing Ecosystem for the Trustworthy Reuse of Healthcare Data

Creating Impact for Rare Diseases

💡 Registries are a powerful instrument to inform policy-making and optimize resources in the healthcare system by delivering comprehensive and accurate data from hospital as important holders of healthcare data. In Belgium, more than 500.000 people are affected by one of over 6.100 known rare diseases, yet fewer than 2% are found in the national rare disease registry (CRRD).

🚀 We launched a pilot project (R3D) in SPECTRE-HD to support hospitals in the identification and diagnosis patients with rare diseases. Our approach leverages existing rare disease markers in the electronic patient records to flag potential new cases for validation by data managers and healthcare practitioners. An important goal is to seamlessly share confirmed rare disease cases with Belgium’s Central Registry of Rare Diseases (CRRD). This benefits the healthcare system and enables more research, and ultimately fosters better outcomes for rare disease patients.

R3D Zebra

Unlock Clinical Notes

💡 We see a great deal of optimism on the potential of new data-driven technologies to transform healthcare. This triggered a surge of interest in the vast amount of unstructured data in electronic patient records. For those outside the field, it’s easy to overlook the significant challenges involved in unlocking this type of data for healthcare innovation.

🚀 Blackbar is a local hospital platform dedicated to the secure, compliant reuse of unstructured documents for advanced analytics and AI. Blackbar lets a hospital automatically create a large pseudonymized corpus of patient documents for reuse in research and innovation projects. Hospitals can even tailor the pseudonymization model to their own patient records, allowing for more control over the risk of direct or indirect patient identification. Finally, Blackbar tackles one of the main concerns in the use of generative AI - patient privacy - since it is also as a platform to safely develop and host LLM applications.

Blackbar

Find and Code Patients

💡 Pseudonymization at scale unlocks free-form clinical narratives for research and innovation. A great example is the CSI web application, that lets a hospital’s clinicians find patients using natural language concepts in clinical documents.

🚀 With CSI, clinicians explore natural language concepts and create groups of them - for example representing a diagnostic marker - with the goal of finding patients. Next to the retrospective identification of patients, clinicians can be notified of new patients that are a match for any concept in a group. CSI will also detect new candidate concepts that appear in new notes. The patients’ disease or condition is often not yet coded in the EHR, so CSI allows the clinician to link natural language concepts to medical codes (e.g., SNOMED CT, ORPHAcodes). These features of CSI, complemented with a REST API, enables a hospital to automatically annotate patients through their existing and new notes, regardless of the EHR system used by the hospital.

CSI

Build a Data Catalog

💡 While data provides valuable insights for your organization, metadata is the key to understanding and managing that data more effectively.

🚀 By publishing a data catalog, everyone in the organization gets access to information on the available datasets and learn about their characteristics. Additional metadata can reveal information about quality and usefulness of available data, so data requesters can assess if the data fits their specific purpose of further processing. Importantly, by aligning from the outset with European standards like DCAT-AP for Health, rich information on health-related data sources can be exchanged and become visible at a regional, national, or European level.

DataHub

Support FAIR Health Data

💡 Making healthcare data findable will inevitably boost its demand. Yet hospitals still struggle with technical and organizational challenges to make their data accessible, interoperable, and reusable — fully FAIR ♻️. Shortages in personnel, expertise, technology, and infrastructure stand in the way of building a data catalog and keeping it up-to-date, managing data requests and delivering data in a timely manner while establishing safe data operations and data management.

🚀 MINE-HD is a health data platform that helps hospitals to close this gap. Based entirely on open-source software, MINE-HD streamlines technology with robust data governance and enables seamless collaboration across teams to meet the growing demand for healthcare data. MINE-HD can be fully deployed on a hospital’s local or cloud-based infrastructure, and lets the data team focus more on areas where they can bring the most value, such as data quality. MINE-HD will evolve with the health data landscape and technology, providing important capabilities for hospitals to collaborate in the European Health Data Space.

MINE-HD

Secure Health Data Processing

💡 Health data holders want legal and safe reuse of data, preserving patient privacy in the entire data-value chain. Health data users need timely, friction-free access to relevant data for their use case, flexibility to work with preferred tools, and adequate storage and compute capacity for their use case. The public must see clear, responsible practices that earn trust in how patient data are reused.

🚀 TREzure is a digital health data platform designed to support the challenges and needs of various stakeholders in the landscape of the European Health Data Space for secondary use. With TREzure, a hospital will grant governed access to multi-modal healthcare data. Built on scalable cloud infrastructure, TREzure features a Secure Processing Environment (SPE) that is operated by a single hospital or a group of hospitals. Leveraging TREzure’s federated data catalog, data users create a “data extraction description” for review by each data holder and for communication to the HDA. These capabilities will significantly shorten the lead time for data users to access the requested data, while data holders can be assured that the entire data-value chain is governed.

TREzure