Integrating AI for Personalized Risk Assessment and Streamlined Clinical Interventions in an Integrated Care System

Prevention of development of comorbidities has been a major focus of healthcare systems. ARISTOTELES’ mission is to develop and validate an innovative AI-based approach embedded within an integrated care system, delivering personalized risk assessments and supporting targeted implementation of management strategies.

The current healthcare landscape mainly relies on epidemiological studies and risk factor control, whereas ARISTOTELES leverages advanced causal machine learning to co-explore disease mechanisms and clinical outcomes in multimorbidity. Traditional approaches focus on established disease associations, often overlooking the intricacies of individual patient profiles. ARISTOTELES pioneers a patient-centred paradigm, offering validated AI tools that consider individual genotypes, phenotypes, lifestyle, occupational/environmental stressors, and socio-economic/behavioural factors.

New central role of AI in healthcare

While in healthcare AI has been mainly used in the backend of clinical informatics, the groundbreaking ambition of ARISTOTELES is to integrate AI into the decision-making process, extending the reach of AI tools beyond physicians to include patients. This shift not only will improve patient awareness of their health issues but will also help them to adopt informed lifestyle changes and proactive interest in their health, leading to ‘living well and ageing well’.

Patients and stakeholders’ engagement

In ARISTOTELES patients and stakeholders will be involved in all stages of AI development, incorporating their views and expertise from the early stages allows patients to express their needs, concerns, and expectations. Patients and their families and caregivers will be placed at the heart of the project to empower them and reduce barriers, promote a culture of dialogue between them and health professionals, foster shared decision-making, and promote health and digital literacy

Atrial fibrillation

Atrial fibrillation (AF) is an irregular and often very rapid heart rhythm (arrhythmia) that can lead to blood clots in the heart, it is the most common type of cardiac arrhythmia worldwide. 1 in 4 adults are at risk of developing AF in their lifetime and the disease is associated with a higher risk of stroke, death, dementia and heart failure. Multimorbidity is higher in patients with AF than in those without, it substantially influences people’s health and quality of life, making management more difficult.

A paradigm shift in AF treatment

By integrating AI into clinical practice, ARISTOTELES will harmonize diverse datasets on AF patients from multiple countries in an agile digital platform. The platform will form a backbone for acceptable, responsible, and respectful uses of patient data to develop and validate novel trustworthy AI tools for predicting personalized risk of various diverse comorbidities.
ARISTOTELES’ goal is to move from a focus on individual risk factors and selected outcomes to a more holistic and integrated approach considering existing or emergent comorbidities. This would timely underpin diagnostic and therapeutic interventions to reduce disease progression, disability, hospitalizations, and all-cause mortality, as well as improve patient adherence to lifestyle modifications, medications, and other treatment regimens.

P4 model of medicine

ARISTOTELES will provide crucial inputs for implementing a continuously learning P4 model of medicine for care of AF patients and associated comorbidities.






  • Develop a platform for data management and analysis, with data harmonized from existing high-quality health relevant data from multiple sources to a common data model.
  • Develop causal, counterfactual ML algorithms and AI solutions for individualised risk prediction, helping patients and healthcare professionals to assess and predict the risk for and/or progression of chronic non-communicable diseases.
  • Facilitate the engagement of patients, their families, and caregivers, using our novel AIs, to adequately monitor their symptoms, seek appropriate care and adopt lifestyle changes.
  • Ease the interaction between patient and physicians. Patients will be better informed to manage their own health, facilitating the management and follow-up through different healthcare providers.
  • At first focusing on patients with AF at risk of multimorbidity, the model is meant to be applied to any complex disease to allow a personalized risk stratification assessment.
  • Assess the needs and regulatory requirements for implementing our novel AIs and understand the potential barriers to the acceptance and adoption of trustworthy AIs across a variety of healthcare systems.


How is our work organized?

The coordinator and management team will oversee the execution of planned activities, the timely submission of deliverables and reports, and the financial and administrative management. WP1 will ensure the effective collaboration among partners, risk mitigation, and successful project implementation.

WP2 aims to develop and implement strategies for ethical, regulatory, and legal compliance across project activities. This WP will create a project-specific trustworthy AI assessment list and monitor progress against its criteria, providing guidance to the Co-ordinator. The goal is to ensure regulatory acceptance of the ARISTOTELES AI system and investigate ethical considerations in its implementation within healthcare settings.

WP3 aims to assess stakeholder understanding, acceptability, and engagement with AI in healthcare, fostering co-production for enhanced usability and dissemination of AI models and platforms. The objectives include establishing stakeholder perspectives, investigating barriers and facilitators of trustworthy AI-based healthcare to co-produce a web-based AI tool interface and evaluating AI tool utilization.

WP4 aims to establish a secure computing infrastructure at the Danish National Genome Center’s high-performance computing centre. It involves integrating datasets from CALIBER, THIN, Danish Health Register, CIPHA, and UK-Biobank onto the platform, multiple datasets from various countries, making the data accessible to WP5 and WP6. The platform, based on a containerized version of eLab, facilitates efficient data management and machine learning development in the medical field. Additionally, a containerized version of the data harmonization pipeline will be created for implementation in WP7.

WP5 will develop novel accurate, reliable and explainable causal AI algorithms for individualised risk prediction new-onset and worsening of comorbidities, along with adverse outcomes. This WP will also develop a web-based tool that implements the AI models, providing an interface for both HCPs and patients.

WP6 will conduct a virtual clinical trial to assess early risk models from WP5 using retrospective cohorts from the available datasets excluded in the training of the AI tool in WP5. The simulation will compare patient management with and without the AI-risk tool, spanning 1 and 5 years, addressing evolving multimorbidity. The generated evidence will be a key source for model refinement and improvement.

WP7 will deliver a cluster-RCT in atrial fibrillation (AF) patients randomised to ‘usual care’ versus adaptative AI-supported patient management, capable of performing in a real-world context. It will compare the outcomes of the RCT with the simulated trial in WP6 and evaluate the implementation strategy and stakeholder views to inform future implementation.

WP8 aims to effectively communicate project ideas, activities, and results to diverse audiences, influencing healthcare policy, patient outcomes, and service delivery. This includes disseminating outputs to healthcare professionals, patients, and caregivers, as well as exploiting results by creating a plan for regulatory acceptability. WP8 will explore alternative routes for implementing scientific advances, ensuring long-term impact. Additionally, it will engage in joint initiatives with other projects, fostering knowledge exchange and clustering activities.