Federated virtual twins for privacy-preserving personalised outcome prediction of type 2 diabetes treatment

Projektinformationen

dAIbetes

ID Finanzhilfevereinbarung: 101136305

DOI

10.3030/101136305

EK-Unterschriftsdatum 14 November 2023

Startdatum 1 Januar 2024

Enddatum 31 Dezember 2028

Finanziert unter

• Health

Gesamtkosten

€ 8 938 145,00

EU-Beitrag

€ 8 938 145,00

Koordiniert durch

UNIVERSITY OF HAMBURG
Germany

Periodic Reporting for period 1 - dAIbetes (Federated virtual twins for privacy-preserving personalised outcome prediction of type 2 diabetes treatment)

Berichtszeitraum: 2024-01-01 bis 2024-12-31

Zusammenfassung vom Kontext und den Gesamtzielen des Projekts

Virtual twins may be used as prognostic tools in precision medicine for personalised disease management. However, their training is a data hungry endeavour requiring big data to be integrated across diverse sources, which in turn is hampered by privacy legislation such as the General Data Protection Regulation. Privacy-enhancing computational techniques, like federated learning, have recently emerged and hold the promise of enabling the effective use of big data while safeguarding sensitive patient information. In dAIbetes, we build on this technology to develop a federated health data platform for clinical application of the first internationally trained federated virtual twin models for diabetes. Our primary medical objective is personalised prediction of treatment outcomes in type 2 diabetes, which afflicts 1 in 10 adults worldwide and causes annual expenditures of ca. 893 billion EUR. While healthcare providers are becoming increasingly effective at targeting diabetes risk factors (e.g. diet or exercises), no guidelines as to the expected outcome for a given treatment for a specific patient exist. To address this urgent, yet unmet need, the federated dAIbetes technology at the heart of which is the federated database network dAIbetes-Net will harmonise existing data of ca. 800,000 type 2 diabetes patients from 6 cohorts distributed across the globe, and learn prognostic virtual twin models, all without requiring the upload of sensitive patient data to a common cloud thus protecting patient privacy. The dAIbetes virtual twin models will be validated for their clinical relevance and applied in clinical practice through the dAIbetes software platform. We aim to demonstrate that our personalised predictions have a prediction error that is at least 10% lower than that of population average-based models. This federated virtual twin technology will enable personalised glycemic control management and act as a blueprint for other complex diseases. Our consortium combines expertise in artificial intelligence, software development, privacy protection, cyber security, and type 2 diabetes and its glycemic control treatment. Ultimately, we aim to resolve the antagonism of privacy and big data in cross-national diabetes research.

Arbeit, die ab Beginn des Projekts bis zum Ende des durch den Bericht erfassten Berichtszeitraums geleistet wurde, und die wichtigsten bis dahin erzielten Ergebnisse

During the first reporting period, dAIbetes has made significant progress toward its overall objectives. The project remains on track according to the timeline established in the grant agreement. We have developed a comprehensive data management plan, and key clinical variables from all participating clinical sites have been identified for training the dAIbetes virtual twin models. To support this effort, we have established a standardized data format, and the clinical sites have begun harmonizing their data accordingly. The development of a standardized database model for the federated database network, dAIbetes-Net, is also progressing as planned, and we are currently working on the necessary and modular app system to ensure efficient data exchange. Additionally, the architectural design of the dAIbetes software platform is moving forward smoothly. The design phase for the dAIbetes virtual twin models has also commenced, representing a crucial step in our project’s development. Overall, the project is supported by a robust governance and decision-making structure, as well as effective project management procedures.

Fortschritte, die über den aktuellen Stand der Technik hinausgehen und voraussichtliche potenzielle Auswirkungen (einschließlich der bis dato erzielten sozioökonomischen Auswirkungen und weiter gefassten gesellschaftlichen Auswirkungen des Projekts)

dAIbetes addresses all four of the expected outcomes as stated in the work programme:

1. Clinicians and other healthcare professionals have access to and/or use validated multi-scale computational models of individual patients for delivering optimised and cost-effective patient management strategies superior to the current standard of care. - The dAIbetes software platform will grant clinicians and other healthcare professionals direct access to the multi-scale computational dAIbetes virtual twin models as part of their clinical practice. The models will have been validated in a feasibility study. Together, this will enable clinicians to offer personalized glycemic treatments for type 2 diabetes, thereby enhancing the overall quality of medical care and ultimately reducing costs linked to ineffective treatment plans.

2. Healthcare professionals benefit from enhanced knowledge of complex disease onset and progression by recourse to validated, multi-scale and multi-organ models. - Type 2 diabetes is an intricate disease influenced by a variety of physiological and environmental factors. Consequently, the current standard of care—which follows a one-size-fits-all treatment approach—is not ideal. Our dAIbetes virtual twin models will provide healthcare professionals with detailed insights into an individual’s disease status and expected progression under specific treatments, thereby facilitating personalized treatment options.

3. Clinicians and patients benefit from new, improved personalised diagnostics, medicinal products, devices, and therapeutic strategies tailored to the individual patient patho-physiology. - Our dAIbetes virtual twin models take into account a wide range of pathophysiological factors to be stored within the federated dAIbetes-Net database network, which will include data from approximately 800,000 subjects. This will allow for highly personalized and optimized care. In the future, the dAIbetes software can be further developed into a medical decision support system, assisting clinicians in streamlining personalized medical care focused on glycermic control for type 2 diabetes patients.

4. Citizens and patients have access to validated ‘virtual twin’ models enabling the integration of citizen-generated data with medical and other longitudinal health data, and benefit from early detection of disease onset, prediction of disease progression and treatment options, and effective disease management. - Alongside the advantages of utilizing the dAIbetes virtual twin models in clinical settings, patients can also benefit in the long-term from continuous incorporation of citizen-generated data for which diabetes is a particular suitable disease entity. The federated technologies that dAIbetes builds on and develops will facilitate a privacy-preserving environment for analysing such personal, sensitive data. The ontology-guided harmonization system and data standards established by dAIbetes allow the integration of information from wearable devices like blood glucose monitors and activity trackers. This comprehensive health data will empower clinicians to identify the onset of the disease and forecast its progression. Additionally, insights from the dAIbetes software provided to patients will support a more personalized and effective approach to disease management.