Inspired by the EU funded project COMFORTAGE (Prediction, Monitoring and Personalized Recommendations for Prevention and Relief of Dementia and Frailty) —coordinated by Dr. George Manias, University of Piraeus, UniPi)— the Interactive Workshop “Patient Digital Twins (PDT) for Dementia and Frailty: Bridging Clinical Practice and AI Technology” brought together clinicians researchers, and patient-centered stakeholders to explore how PDT technology can support clinical practice and improve patient well-being.
The Workshop fostered dialogue between clinical and technical communities examining dementia and frailty care from both traditional and AI-driven perspectives. Dr Eva Ntanasi provided the clinical lead, while Stelios Kokkas (CERTH) provided the technical lead, and Dr Sofia Segkouli coordinated the discussion. Panelists included Sokratis Papageorgiou, Professor of Neurology, NKUA; Elias Panagiotopoulos, Emeritus Professor, University of Patras, Orthopaedic Surgeon; Christina Malakou, Cardiologist-DEPAN; and Paraskevi Sakka, Neurologist-Psychiatrist, Alzheimer Athens.
Workshop Methodology & Objectives
A distinctive feature of the workshop was its parallel examination methodology. Participants were presented with synthetic but realistic patient case studies specifically designed to illustrate different aspects of dementia and frailty progression. For each case, clinicians demonstrated their standard assessment procedures and multidisciplinary care planning. For the next step, the same patient profiles were analyzed using PDT models, which generated predictive insights regarding disease progression, risk trajectories, and potential intervention strategies. This side-by-side comparison created an interactive environment, enabling participants to directly compare clinical reasoning with computational prediction and to reflect on complementarities, discrepancies, and opportunities for integration.
The workshop pursued several key objectives. These included demonstrating current best practices in dementia and frailty assessment, presenting the predictive capabilities of PDT technology, identifying synergies and gaps between clinical intuition and AI outputs, gathering stakeholder views on barriers and opportunities for adoption, and fostering meaningful collaboration between healthcare professionals and AI developers. To encourage engagement and collective reflection, an interactive Mentimeter session was conducted, allowing participants to anonymously share perspectives, concerns, and expectations regarding PDT implementation.
Digital Twins in Dementia Care: The User Perspective
Discussion was initiated by Dr. Sofia Segkouli, whose introductory presentation, “Digital Twins in Dementia Care: The User Perspective,” provided a conceptual framework for the dialogue. Digital twins were defined as dynamic, continuously updated virtual replicas of patients that integrate data from clinical assessments, wearable technologies, neuroimaging, and potentially genomics. In dementia care, this longitudinal and multidimensional integration was recognized as particularly valuable. Clinicians emphasized the persistent difficulty of achieving timely and accurate diagnoses. Through the comparative case analyses, participants observed that digital twins can identify subtle longitudinal patterns that may not be immediately evident in cross-sectional clinical encounters. The workshop concluded that PDT systems may enhance diagnostic precision, by supporting differential diagnosis and early detection.
From reactive to predictive care in dementia
A central theme of the discussion was the shift from reactive to predictive care. Traditional dementia management often responds to crises, rather than anticipating them. In contrast, PDT models continuously analyze data to forecast risk and simulate intervention outcomes before decline becomes evident. This predictive approach supports precision medicine and enables proactive, individualized strategies. Participants agreed it could optimize pharmacological and non-pharmacological interventions, improving timing and potentially slowing disease progression.
Ethical Issues & Data Governance
However, personalization was recognized as both an opportunity and a challenge. Representing a person through a digital simulation requires accurate data and sustained respect for autonomy, consent, and dignity. As decisional capacity in dementia may decline, maintaining meaningful and ongoing consent is difficult. The key ethical issue is not only data ownership, but whether individuals retain genuine control over how their data and predictive outputs are used. Without transparency and mechanisms for revising or withdrawing consent, there is a risk that digital twins could shift from tools of representation to instruments of surveillance. Closely related to autonomy is data governance. Participants agreed that responsible implementation demands comprehensive oversight across the entire data lifecycle, including collection, storage, integration, security, and risk assessment for re-identification. Governance was framed not as a purely technical issue, but as a shared responsibility requiring collaboration among clinicians, AI researchers, ethicists, and legal experts. Bias and fairness also emerged as significant concerns. If digital twins are trained on non-representative datasets, they may inadvertently reinforce disparities related to age, ethnicity, socioeconomic status, or disability. Participants stressed the need for inclusive data practices, participatory design involving patients and caregivers, and continuous auditing of algorithmic outputs to ensure fairness.
Clinical Trust – The roles of HCPs and Patients
Transparency and explainability were highlighted as prerequisites for clinical trust. Many predictive models function as “black boxes,” generating recommendations without clarifying their reasoning, which make clinicians hesitant to rely on them. Explainable PDT models, clarifying which variables influenced predictions and how risk factors were weighted, were considered essential for safe integration into clinical workflows.
The role of healthcare professionals was another focal point. Participants agreed that PDTs will not replace clinicians but will reshape their responsibilities. The risk of automation bias over reliance on algorithmic outputs was acknowledged. Clinicians will increasingly serve as interpreters, translating predictive insights into patient-centered care while applying ethical judgment and contextual awareness.
From the patient perspective, empowerment emerged as central. PDTs can enhance shared decision-making by clarifying disease progression and intervention options. However, meaningful empowerment requires accessibility, cultural sensitivity, and usability for individuals with varying levels of digital literacy and cognitive capacity.
