Periodic Reporting for period 1 - TARGET (Health virtual twins for the personalised management of stroke related to atrial fibrillation)
Período documentado: 2024-01-01 hasta 2025-06-30
Atrial Fibrillation (AF) is the most common sustained cardiac arrhythmia worldwide, and its complications represent a major public health burden. Patients with AF face a 5-fold increased risk of ischaemic stroke, and AF-related stroke (AFRS) is often more severe than strokes of other origins, leading to higher mortality, poorer functional recovery, long-term disability, and reduced quality of life for patients and caregivers. Despite extensive research and advances in stroke prevention, major challenges remain in understanding the complex links between AF and stroke, tailoring treatment to individual patients, and managing long-term risks such as stroke recurrence or bleeding complications. These challenges are compounded by an ageing population, rising multimorbidity, and increasing pressures on healthcare systems across Europe.
TARGET addresses these unmet needs by developing novel virtual twin-based AI models that combine mechanistic and data-driven approaches with causal AI. These models aim to bridge the gap between research and clinical practice, enabling personalised care at every stage of the AF-related stroke pathway: from prevention and early diagnosis, to acute treatment optimisation and post-stroke rehabilitation. TARGET’s virtual twins will provide dynamic, individualised risk and outcome predictions, offering clinicians and patients new tools to guide decision-making.
TARGET’s objectives:
- Develop personalised risk prediction models for AF and AFRS, grounded in causal understanding of underlying pathophysiological processes and novel biomarkers.
- Optimise treatment and rehabilitation through virtual twin models that can dynamically assess and predict outcomes, tailoring strategies to individual needs.
- Advance causal inference in clinical medicine by elucidating disease drivers and their impact on AF progression and stroke complications.
- Accelerate translational research through multi-scale, multi-organ modelling of the heart, brain, and neuromusculoskeletal system, integrated with real-world clinical and imaging data.
- Develop interoperable decision support tools and a secure data integration and sharing platform to facilitate clinical adoption and sustainability.
- Evaluate facilitators and barriers to implementation of these tools, ensuring that usability, trust, and acceptability among patients and healthcare professionals are embedded from the outset.
TARGET aims to generate impact at multiple levels:
- Scientific impact: advancing virtual twin and causal AI methodologies for personalised medicine, producing new knowledge on the causal mechanisms of AF and AFRS, and identifying novel biomarkers for risk prediction and outcomes.
- Technological impact: delivering secure, and ethically sound decision-support tools and digital health solutions at TRL6, strengthening Europe’s leadership in health technology and digital innovation.
- Societal impact: empowering patients and caregivers with personalised, trustworthy tools; improving quality of life through better prevention, acute management, and rehabilitation; and fostering trust in digital health technologies through co-creation and transparent design.
- Economic impact: reducing the direct and indirect costs of stroke care in Europe and worldwide, through fewer hospitalisations, improved rehabilitation outcomes, and more efficient resource allocation, while building pathways for long-term exploitation and scalability.
TARGET addresses these unmet needs by developing novel virtual twin-based AI models that combine mechanistic and data-driven approaches with causal AI. These models aim to bridge the gap between research and clinical practice, enabling personalised care at every stage of the AF-related stroke pathway: from prevention and early diagnosis, to acute treatment optimisation and post-stroke rehabilitation. TARGET’s virtual twins will provide dynamic, individualised risk and outcome predictions, offering clinicians and patients new tools to guide decision-making.
TARGET’s objectives:
- Develop personalised risk prediction models for AF and AFRS, grounded in causal understanding of underlying pathophysiological processes and novel biomarkers.
- Optimise treatment and rehabilitation through virtual twin models that can dynamically assess and predict outcomes, tailoring strategies to individual needs.
- Advance causal inference in clinical medicine by elucidating disease drivers and their impact on AF progression and stroke complications.
- Accelerate translational research through multi-scale, multi-organ modelling of the heart, brain, and neuromusculoskeletal system, integrated with real-world clinical and imaging data.
- Develop interoperable decision support tools and a secure data integration and sharing platform to facilitate clinical adoption and sustainability.
- Evaluate facilitators and barriers to implementation of these tools, ensuring that usability, trust, and acceptability among patients and healthcare professionals are embedded from the outset.
TARGET aims to generate impact at multiple levels:
- Scientific impact: advancing virtual twin and causal AI methodologies for personalised medicine, producing new knowledge on the causal mechanisms of AF and AFRS, and identifying novel biomarkers for risk prediction and outcomes.
- Technological impact: delivering secure, and ethically sound decision-support tools and digital health solutions at TRL6, strengthening Europe’s leadership in health technology and digital innovation.
- Societal impact: empowering patients and caregivers with personalised, trustworthy tools; improving quality of life through better prevention, acute management, and rehabilitation; and fostering trust in digital health technologies through co-creation and transparent design.
- Economic impact: reducing the direct and indirect costs of stroke care in Europe and worldwide, through fewer hospitalisations, improved rehabilitation outcomes, and more efficient resource allocation, while building pathways for long-term exploitation and scalability.
In 18 months, TARGET has made strong progress across all Specific Objectives (SOs) and remains on track to meet milestones.
• SO1 Stakeholder engagement: Ethical approvals secured; participant identification started in UK, ES, BE, NL. Preparations underway for co-production workshops with patients and HCPs.
• SO2 Digital twins: Initial multi-scale, multi-organ virtual twins of heart, brain, NMS developed, with early TARGET data integration. Basis for refinement toward TRL6.
• SO3 Risk prediction: Baseline AI and causal AI models built using large clinical datasets; DL models for AF/arrhythmia detection developed. Interpretability methods embedded; models ready for integration into risk-prediction tools.
• SO4 Diagnosis & management: First AFRS diagnosis and early management models created with retrospective data. Multi-modal architectures handling incomplete data in development and prepared for VT integration.
• SO5 Rehabilitation: AI models predicting functional outcomes from clinical and imaging data show strong performance. Clustering for rehab needs and prototype recommendation engine for personalised therapy in progress.
• SO6 Clinical studies: Four observational studies (NOTE-AF UK; TAILOR ES, NL; PEARL ES; FOSTER BE) covering the full AF–stroke pathway have begun recruitment and are generating initial datasets for validation and in-silico trials.
• SO7 Decision support & platform: Prototypes of 5 tools in progress: dynamic risk tool, real-time monitoring, stroke management, rehab programmes, and serious-games rehab. GDPR-compliant data integration/sharing platform operational, ensuring interoperability and secure use.
• SO1 Stakeholder engagement: Ethical approvals secured; participant identification started in UK, ES, BE, NL. Preparations underway for co-production workshops with patients and HCPs.
• SO2 Digital twins: Initial multi-scale, multi-organ virtual twins of heart, brain, NMS developed, with early TARGET data integration. Basis for refinement toward TRL6.
• SO3 Risk prediction: Baseline AI and causal AI models built using large clinical datasets; DL models for AF/arrhythmia detection developed. Interpretability methods embedded; models ready for integration into risk-prediction tools.
• SO4 Diagnosis & management: First AFRS diagnosis and early management models created with retrospective data. Multi-modal architectures handling incomplete data in development and prepared for VT integration.
• SO5 Rehabilitation: AI models predicting functional outcomes from clinical and imaging data show strong performance. Clustering for rehab needs and prototype recommendation engine for personalised therapy in progress.
• SO6 Clinical studies: Four observational studies (NOTE-AF UK; TAILOR ES, NL; PEARL ES; FOSTER BE) covering the full AF–stroke pathway have begun recruitment and are generating initial datasets for validation and in-silico trials.
• SO7 Decision support & platform: Prototypes of 5 tools in progress: dynamic risk tool, real-time monitoring, stroke management, rehab programmes, and serious-games rehab. GDPR-compliant data integration/sharing platform operational, ensuring interoperability and secure use.
By Month 18, TARGET has published 13 peer-reviewed papers across cardiology, stroke, digital health, and AI/ML, underscoring the project’s strong interdisciplinary character.
A central focus has been the advancement of AI and virtual twin modelling along the AF-stroke pathway. Publications have mapped how AI and digital twin technologies can support prevention, acute care, and rehabilitation. The consortium has contributed new insights into AF in critical illness, introduced TARGET’s clinical studies and their objectives, and proposed innovative AI methods for AF phenotyping and patient stratification. Further work extended into rehabilitation, using muscle synergy analysis in stroke survivors to inform modelling approaches. Together, these outputs demonstrate TARGET’s leadership in developing clinically relevant models for AF and stroke. Complementing this, the project has delivered key contributions in semantic interoperability, data quality, and knowledge representation. Publications presented methods for schema languages to improve interoperability, synthetic health knowledge graph generation, and a novel framework for semantic exploration of FHIR-compliant electronic health records. These advances lay the technical foundations for scalable, high-quality data use in digital twin applications.
TARGET partners have also shaped expert consensus and clinical recommendations, covering AF management in older populations and a European Society of Cardiology consensus on measuring AF burden and the use of digital tools, relevant to clinical practice and the integration of technology in care pathways. The consortium has addressed the broader health economics and policy context of AF and stroke by discussing regional and global challenges in AF management. These contributions highlight TARGET’s capacity to contribute to technological and clinical developments, as well as policy-level decision-making.
This body of work illustrates how TARGET is advancing knowledge across clinical, technical, and policy dimensions, firmly positioning the project as a leader in AI-driven digital twin research for cardiovascular disease.
A central focus has been the advancement of AI and virtual twin modelling along the AF-stroke pathway. Publications have mapped how AI and digital twin technologies can support prevention, acute care, and rehabilitation. The consortium has contributed new insights into AF in critical illness, introduced TARGET’s clinical studies and their objectives, and proposed innovative AI methods for AF phenotyping and patient stratification. Further work extended into rehabilitation, using muscle synergy analysis in stroke survivors to inform modelling approaches. Together, these outputs demonstrate TARGET’s leadership in developing clinically relevant models for AF and stroke. Complementing this, the project has delivered key contributions in semantic interoperability, data quality, and knowledge representation. Publications presented methods for schema languages to improve interoperability, synthetic health knowledge graph generation, and a novel framework for semantic exploration of FHIR-compliant electronic health records. These advances lay the technical foundations for scalable, high-quality data use in digital twin applications.
TARGET partners have also shaped expert consensus and clinical recommendations, covering AF management in older populations and a European Society of Cardiology consensus on measuring AF burden and the use of digital tools, relevant to clinical practice and the integration of technology in care pathways. The consortium has addressed the broader health economics and policy context of AF and stroke by discussing regional and global challenges in AF management. These contributions highlight TARGET’s capacity to contribute to technological and clinical developments, as well as policy-level decision-making.
This body of work illustrates how TARGET is advancing knowledge across clinical, technical, and policy dimensions, firmly positioning the project as a leader in AI-driven digital twin research for cardiovascular disease.