Building TrustworthyDigital Twins for Patient-Centred Medicine
Work Packages
Click on the work package areas to learn more about each team’s role in the TWIN-X project
Work Package 1 – Project management and scientific coordination
WP1 establishes efficient management and communication structures to ensure the successful implementation of the project. It provides scientific and administrative coordination of all project activities and integrates proactive risk assessment and mitigation measures to ensure the timely delivery, high quality, and impact of project results.
Work Package 2 – Representations from unstructured clinical data
WP2 develops methods to transform clinical text and unstructured patient data into harmonised, research-ready information. It creates a common framework for organising data, extracts key clinical information and events, and combines them into longitudinal patient timelines. Furthermore, the partners involved in this work package seek to create AI models that can analyse and reason across different types of clinical data while providing transparent and traceable results. The resulting patient representations and evaluation methods will support further research and healthcare applications.
Work Package 3 – Medical imaging and histopathology embeddings
The development of innovative AI methods for analysing medical imaging data from radiology and pathology lies at the core of WP3. It creates image embeddings that capture relevant patterns across different image types, sites, vendors, and scanners. Within the WP, the research team also establishes quality control procedures, confidence scoring, and cross-modality calibration to ensure reliable and generalisable results.
Work Package 4 – Representations from social determinants and environmental exposures
WP4 provides methods to integrate social, environmental, and demographic factors into the project’s data framework. The research team establishes governance and ethical guidelines for data linkage, creates variables and representations of social determinants and environmental exposures, and assesses potential biases and fairness issues. The WP delivers harmonised data that can be combined with other project data sources for further analysis.
Work Package 5 – Integrated Digital Twin Framework
WP5 focuses on bringing together the information generated across different data modalities into a unified framework. To this end, the research team harmonises modality-specific representations, develops mechanisms for information retrieval and reasoning, and enables users to interact with the system through clinically relevant queries. The WP also provides interfaces and reference implementations to facilitate the use of the developed methods in future research.
Work Package 6 – Clinical demonstrators and validation
WP6 demonstrates and evaluates the practical application of the digital twin in the fields of oncology and cardiology. Based on clearly defined protocols, endpoints, and trust criteria, the clinical team develops and iteratively assesses two demonstrators. Particular emphasis is placed on validating robustness, reliability, and generalisability across different clinical sites and settings.
Work Package 7 – Trustworthy and explainable AI
WP7 focuses on ensuring the trustworthiness of the AI models developed within the project. The research team will advance methods for model calibration, uncertainty quantification, and distribution shift detection to improve the reliability and robustness of AI predictions. In addition, the team will enhance the explainability of AI models across imaging, pathology, text, and reasoning tasks. The work package will also address the identification, assessment, and mitigation of potential biases, including those related to sex and gender to promote fairness and equitable model performance.
Work Package 8 – Dissemination, exploitation, open science and training
WP8 is dedicated to increasing the project’s visibility and raising awareness of the development and application of digital twins in biomedicine and foster the sustainable exploitation of TWIN-X’s project results. The work package aims to build and strengthen networks with key stakeholders from academia, industry, healthcare, and policy-making. Through targeted communication activities, systematic stakeholder analysis, and proactive engagement strategies, WP8 will ensure that project results are accessible to diverse audiences. WP8 ensures the sustainable exploitation of project results and facilitates their translation into clinical practice. To this end, WP8 will also develop and deliver a dedicated training programme for both clinical and technical audiences, contributing to capacity building and skills development in the field of medical AI and digital twin technologies across Europe.